Tue, 22 Jun 2021 10:01:43 CDT | login

Information for build frontier-squid-4.15-2.1.osg36.el7

Package Namefrontier-squid
SummaryThe Frontier distribution of the Squid proxy caching server
DescriptionSquid is a high-performance proxy caching server for Web clients, supporting FTP, gopher, and HTTP data objects. Unlike traditional caching software, Squid handles all requests in a single, non-blocking, I/O-driven process. Squid keeps meta data and especially hot objects cached in RAM, caches DNS lookups, supports non-blocking DNS lookups, and implements negative caching of failed requests. This squid distribution has been expecially tuned to work with the Frontier server to cache Database entries retrieved from an ORACLE back-end, particularly its need for the http standard If-Modified-Since feature.
Built byDave Dykstra
State complete
StartedWed, 09 Jun 2021 21:02:11 CDT
CompletedWed, 09 Jun 2021 21:10:20 CDT
Taskbuild (osg-3.6-el7, /svn/native/redhat/branches/osg-3.6/frontier-squid:26501)
Extra{'source': {'original_url': 'svn+https://vdt.cs.wisc.edu/svn/native/redhat/branches/osg-3.6/frontier-squid#26501'}}
frontier-squid-4.15-2.1.osg36.el7.src.rpm (info) (download)
frontier-squid-4.15-2.1.osg36.el7.x86_64.rpm (info) (download)
Changelog * Fri Jun 04 2021 Dave Dykstra <dwd@fnal.gov> 4.15-2.1 - Upgrade to 4.15-2 tarball with the following release notes: - Fix SQUID_COMPRESS_LOGS=false, which was broken since 4.13-3. Change it to also use the external logrotate command instead of the squid builtin log rotation. * Tue May 11 2021 Dave Dykstra <dwd@fnal.gov> 4.15-1.2 - Filter out unneeded python3 dependency with the __requires_exclude_from macro. * Mon May 10 2021 Dave Dykstra <dwd@fnal.gov> 4.15-1.1 - Upgrade to 4.15-1 tarball with the following release notes: - Update to squid-4.15 with release announcement at http://lists.squid-cache.org/pipermail/squid-announce/2021-May/000127.html The update includes security fixes, and at least one (SQUID-2020:11) is relevant because it enables someone that is allowed by access controls to use squid for some purpose to bypass any other access controls and use it for any purpose. That one can be worked around with "uri_whitespace deny", but other denial of service vulnerabilities fixed in this version have no workaround. - Set "uri_whitespace deny" by default to avoid any other related vulnerabilities in the future. - Exit with an error if either of the two required logfile_rotate entries are not in squid.conf. This is for the protection of who are using SQUID_CUSTOMIZE=false and have not properly merged the changes from squid.conf.frontierdefault. - Update the customhelps.awk function setoption() to work correctly after a setoptionparameter(). - Allow cache_dir null type which means no disk cache, for systems that have very large cache_mem and maximum_object_size_in_memory. * Thu Apr 22 2021 Dave Dykstra <dwd@fnal.gov> 4.13-5.3 - Change to use systemctl instead of service commands where possible. * Tue Mar 30 2021 Dave Dykstra <dwd@fnal.gov> 4.13-5.2 - Fix bad permissions on one of the shoal-agent dependency directories. Only caused a problem when rebuilding from src rpm. * Mon Mar 29 2021 Dave Dykstra <dwd@fnal.gov> 4.13-5.1 - Upgrade to 4.13-5 tarball with the following release notes: - Use /bin/bash instead of $SHELL in fn-local-squid.sh because $SHELL is sometimes set to /bin/nologin (in particular when the squid user directly invokes fn-local-squid.sh). This affected cleaning out old cache files and starting shoal. - Do not attempt to set net.core.somaxconn higher than the kernel maximum of 65335. * Fri Mar 19 2021 Dave Dykstra <dwd@fnal.gov> 4.13-4.1 - Increase shoal-agent version to 1.0.0. - Upgrade to 4.13-4 tarball with the following release notes: - Add Compute Canada stratum 1s to MAJOR_CVMFS - Add any parameters passed in by $SQUID_START_ARGS to the starting of squid. This is set to --foreground by the OSG frontier-squid docker container. * Thu Mar 18 2021 Dave Dykstra <dwd@fnal.gov> 4.13-3.2 - Fix the building of shoal-agent to be able to work again with no network access (for OSG koji). * Wed Mar 17 2021 Dave Dykstra <dwd@fnal.gov> 4.13-3.1 - Upgrade to 4.13-3 tarball with the following release notes: - Change log rotates to no longer use reconfigure signal but to instead set logfile_rotate to zero and use the rotate signal. This is much lighter weight and works around squid bug #5113. Administrators can still set logfile_rotate to non-default values, but it is used only by fn-local-squid.sh and overridden for squid. - With multiple squid services, send the rotate signal only once to each squid. - Add timestamps to squidcron.log for the beginning and end of rotation. - Remove el6 support * Mon Sep 28 2020 Dave Dykstra <dwd@fnal.gov> 4.13-2.1 - Upgrade to 4.13-2 tarball with the following release note: - Move temporary logrotate config file to the log directory, to keep SELinux on CentOS8 happy. * Mon Jun 29 2020 Dave Dykstra <dwd@fnal.gov> 4.13-1.1 - Upgrade to 4.13-1 tarball with the following release notes: - Update to squid-4.13 with release announcement at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00117.html It includes a couple of relevant security advisories related to cache poisoning. - Remove patch for bug 5051 since it is included in the 4.13 release. - Remove the recursion on the restorecon for SELinux in the %post install step, to avoid taking a long time when going through a large cache. * Mon Jun 29 2020 Dave Dykstra <dwd@fnal.gov> 4.12-2.1 - Upgrade to 4.12-2 tarball with the following release notes (from 4.12-1 and 4.12-2): - Update to squid-4.12 with release announcement at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00114.html It includes a couple of security advisories, but they are on features that as far as I know are not used by frontier-squid users. - Remove patches for bug #5030 and #5041 which are included in 4.12 - Slightly update patch for bug #5051 as recommended by squid consultant - Note that an additional, more relevant squid advisory fixed in 4.12 was published late: https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00116.html - Support the "stdio:" prefix for access_log * Thu Jun 11 2020 Dave Dykstra <dwd@fnal.gov> 4.11-4.3 - Require systemd-devel on non-el6 systems, so squid's configure will recognize that it is on a systemd-based system. - Update to shoal-agent-0.9.9, and remove python3 patch because that is now merged. * Thu Jun 11 2020 Dave Dykstra <dwd@fnal.gov> 4.11-4.2 - Fix quoting on el6 build options to select the correct types of basic authentication. The bug accidentally enabled ldap authentication which prevented an el6 rpm from installing on el7. - Remove openldap-devel build dependency. * Tue Jun 09 2020 Dave Dykstra <dwd@fnal.gov> 4.11-4.1 - Upgrade to 4.11-4 tarball with the following release note: - Apply patch for bug #5041 which broke compilation in squid-4.11 on el7 with systemd. - Update rpm spec to build on el8. - Use python3 for shoal instead of python2 because that has the greatest portability across operating systems. On el6, use the scl add-on package from Red Hat. Apply a patch to shoal to work with python3. - Pre-download python pip packages into shoal-downloads.tar.gz to include them in the src rpm. - Disable NIS basic-authentication on el8 because required header files are missing * Wed Jun 03 2020 Dave Dykstra <dwd@fnal.gov> 4.11-3.1 - Upgrade to 4.11-3 tarball with the following release notes: - Apply patch for bug #5051 which prevents a negative cache from persisting indefinitely with if-modified-since and collapsed forwarding. - Fix shoal when there are multiple squid workers. - Add cc.*\.in2p3\.fr to MAJOR_CVMFS. It was already in ATLAS_FRONTIER so it was included for installations that accept both, but not for those that accept only MAJOR_CVMFS. * Thu Apr 23 2020 Dave Dykstra <dwd@fnal.gov> 4.11-2.1 - Upgrade to 4.11-2 tarball with the following release notes: - The 4.11 release announcement is now at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00108.html The announcement was delayed because security vulnerabilities were being double-checked. - frontier-squid was not susceptible to the ESI vulnerability, because of missing libraries on the build machine. Add --disable-esi to make sure it doesn't get accidentally enabled. * Tue Apr 21 2020 Dave Dykstra <dwd@fnal.gov> 4.11-1.1 - Upgrade to 4.11-1 tarball with the following release notes: - Update to squid-4.11. There's no announcement yet but the change log is at http://www.squid-cache.org/Versions/v4/changesets/. - Remove patches for bugs #5022 and #5036 which are in version 4.11. The patch for #5030 is still applied. * Mon Apr 13 2020 Dave Dykstra <dwd@fnal.gov> 4.10-4.1 - Upgrade to 4.10-4 tarball with the following release note: - Add patch for bug 5036 which caused a varying number of capital 'L's to appear at the beginning of access.log lines when the log buffer overflows. * Mon Mar 16 2020 Dave Dykstra <dwd@fnal.gov> 4.10-3.1 - Upgrade to 4.10-3 tarball with the following release notes: - Update patch for squid bug #5022 to final version. - Add patch for squid bug #5030 which reported that negative caching was not working. This is an important feature for keeping load down on CVMFS stratum 1s that are not hosting a repository. * Thu Feb 13 2020 Dave Dykstra <dwd@fnal.gov> 4.10-2.1 - Upgrade to 4.10-2 tarball with the following release notes: - Apply patch for squid bug #5022, to prevent a reconfigure from crashing the coordinator process when there are multiple workers. - Use reconfigure signal for log rotation in all cases again. * Mon Feb 03 2020 Dave Dykstra <dwd@fnal.gov> 4.10-1.1 - Upgrade to 4.10-1 tarball with the following release notes: - Update to squid-4.10, with release notes at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00103.html including a fix for a serious security vulnerability affecting reverse proxies and a potential information leak when proxying ftp. - Remove patch for bug #4735, as it has been included in the 4.10 release. - Change compressing log rotation back to use copytruncate when there are multiple workers, because the reconfigure signal triggers an SNMP bug when there are multiple workers (squid bug #5022). - Disable log rotatation when SQUID_MAX_ACCESS_LOG=0. Using this is highly discouraged since standard logrotate is usually unable to keep up with the high volume of logs typically generated by squid. It also interferes with frontier-awstats support. - Change log rotating cron scripts to ignore commented lines in /etc/sysconfig/frontier-squid * Mon Jan 27 2020 Dave Dykstra <dwd@fnal.gov> 4.9-5.1 - Upgrade to 4.9-4 tarball with the following release note: - Change compressing log rotation to not use logrotate's copytruncate. Instead, use create and a postrotate that sends squid a reconfigure signal. This is needed when using rsyslog's imfile module to copy to syslog, but also greatly speeds up rotation when log files are large. * Thu Jan 02 2020 Dave Dykstra <dwd@fnal.gov> 4.9-4.1 - Upgrade to 4.9-4 tarball with the following release note: - Limit the hard nofile ulimit check to the "start" operation. * Tue Dec 31 2019 Dave Dykstra <dwd@fnal.gov> 4.9-3.1 - Upgrade to 4.9-3 tarball with the following release notes: - Apply a patch for bug #4735, to prevent caching an object when its http/1.1 last chunk (that is, zero-length chunk) is missing. - Disable the URN proto by default, to avoid potential future security vulnerabilities. - Fail if the hard nofile ulimit for the squid user is less than 4096 (which is unlikely since it is the default value on RHEL6 & 7) and increase the soft nofile ulimit to match the hard limit. Squid uses the soft limit as the default max_filedescriptors. Previously the root portion of the startup script attempted to increase the soft limit to 4096, but it turned out to have no effect because runuser resets it to its default for that user. * Sun Nov 10 2019 Dave Dykstra <dwd@fnal.gov> 4.9-2.1 - Upgrade to 4.9-2 tarball with the following release note: - Fix bug in daily cron that prevented it from rotating any logs * Fri Nov 08 2019 Dave Dykstra <dwd@fnal.gov> 4.9-1.1 - Upgrade to 4.9-1 tarball with the following release notes: - Update to squid-4.9 with release notes at http://squid-web-proxy-cache.1019090.n4.nabble.com/squid-announce-Squid-4-9-is-available-td4688506.html - Add cvmfs01.nikhef.nl, cvmfs-stratum-one.zeuthen.desy.de, and grid-cvmfs-one.desy.de to the MAJOR_CVMFS acl. * Thu Aug 29 2019 Dave Dykstra <dwd@fnal.gov> 4.8-2.1 - Upgrade to 4.8-2 tarball with the following release note: - Add support for starting and stopping shoal-agent when installed and enabled by SQUID_AUTO_DISCOVER=true - Add compilation and installation for shoal-agent * Wed Jul 17 2019 Dave Dykstra <dwd@fnal.gov> 4.8-1.1 - Upgrade to 4.8-1 tarball with the following release notes: - Update to squid-4.8 with release notes at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00096.html There was no 4.7 announcement but here's the ChangeLog: https://github.com/squid-cache/squid/blob/f977bfa698ab92e9474c775cef0e01fb756a4b0f/ChangeLog * Wed Jul 17 2019 Dave Dykstra <dwd@fnal.gov> 4.4-2.1 - Upgrade the 4.4-1.1 version to the 4.4-2 tarball with the following release note: - Apply security patch for CVE-2019-12527 http://www.squid-cache.org/Advisories/SQUID-2019_5.txt which can (before being patched) be exploited through the ftp protocol. * Fri May 17 2019 Dave Dykstra <dwd@fnal.gov> 4.6-2.1 - Upgrade to 4.6-2 tarball with the following release notes: - Add support for sending access_log through syslog - Add additional allowed monitoring host CERN IP range of * Mon Apr 08 2019 Dave Dykstra <dwd@fnal.gov> 4.6-1.1 - Upgrade to 4.6-1 tarball with the following release notes: - Update to squid-4.6 with release notes at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00090.html squid-4.5 release notes are at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00089.html - If a rotate is blocked because processing awstats logs is running long, do the rotate during the next rotateiflarge cron instead. * Wed Oct 31 2018 Dave Dykstra <dwd@fnal.gov> 4.4-1.1 - Upgrade to 4.4-1 tarball with the following release notes: - Update to squid-4.4 with release notes at https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00086.html including one fairly significant security fix for a potential denial of service due to memory leaks on rejected SNMP queries. - If different squid services have different numbers of workers (for example by using setserviceoption on "workers") then use the biggest number when creating cache and log directories. * Thu Oct 04 2018 Dave Dykstra <dwd@fnal.gov> 4.3-1.1 - Upgrade to 4.3-1 tarball with the following release notes: - Major update to squid-4.3. - Removed patch for bug 4616 (patch was already applied in 4.3). - Replaced patch for bug 3952 with new --foreground option. - Ported patch for bug 7 forward. - Compile with devtoolset-2-toolchain on el6, in order to use c++11 - Add PAM basic authentication module - Disable security-cert-validators to avoid perl Crypt::OpenSSL:X509 dependency * Tue Aug 28 2018 Dave Dykstra <dwd@fnal.gov> 3.5.28-2.1 - Upgrade to 3.5.28-2 tarball with the following release notes: - Set umask and PATH in fn-local-squid.sh because /etc/init.d/functions used to set them. This fixes problems with log rotation from cron. - Invoke /etc/init.d/functions from /etc/init.d/frontier-squid if it exists, because that is how systemd-based systems invoke systemctl to keep track of the service. If it does not exist, set the PATH. * Thu Aug 23 2018 Dave Dykstra <dwd@fnal.gov> 3.5.28-1.1 - Upgrade to 3.5.28-1 tarball with the following release notes: - Update to squid-3.5.28. Here is the release announcement: https://www.mail-archive.com/squid-announce@lists.squid-cache.org/msg00083.html Removed patches for advisories SQUID-2018_1 and SQUID-2018_2 and for bug 4767 because fixes are included in the release. - Remove including /etc/init.d/functions from fn-local-squid.sh and from /etc/init.d/frontier-squid. It wasn't being used and isn't in some docker containers such as slc6-lite. - If there are *.log.0.gz files in the presence of awstats, remove them before rotating. They prevented rotation from working on el7. - When creating an account for running squid under, use a home directory of /var/lib/squid instead of defaulting to /home/${FRONTIER_USER:-squid}. The old default doesn't work on systems with automounted /home. * Thu Jul 19 2018 Dave Dykstra <dwd@fnal.gov> 3.5.27-5.2 - Add %verify (not user group) to the %ghost statements for cache dir and log dir, and add %verify (not user group mode) on squid.conf and squid.conf.old. Without those, rpm -V complains on el7 only, not el6. * Fri Jul 06 2018 Dave Dykstra <dwd@fnal.gov> 3.5.27-5.1 - Upgrade to 3.5.27-5 tarball with the following release note: - Avoid a reverse DNS lookup for every client connection by using the workaround in https://bugs.squid-cache.org/show_bug.cgi?id=4575, which is to override the defaults for url_rewrite_extras and store_id_extras. * Wed Apr 18 2018 Dave Dykstra <dwd@fnal.gov> 3.5.27-4.2 - Support SELinux by default by doing resorecon -R /var/cache/squid during postinstall. * Thu Mar 01 2018 Dave Dykstra <dwd@fnal.gov> 3.5.27-4.1 - Upgrade to 3.5.27-4 tarball with the following release note: - If the maximum listen backlog (sysctl net.core.somaxconn) is less than 1/4th the file descriptor limit (ulimit -n) when starting squid, increase it to that amount. That's the listen backlog that squid requests, and the larger backlog helps when a server is hit so hard with requests that squid can't keep up. * Wed Jan 24 2018 Dave Dykstra <dwd@fnal.gov> 3.5.27-3.1 - Upgrade to 3.5.27-3 tarball with the following release notes: - Add configuration options to always honor "Pragma: no-cache" from a client, even if the client also sends a "Cache-control" header, as the current frontier-client (v2.8.20) always does. https://bugs.squid-cache.org/show_bug.cgi?id=4809 - Included provided patches from the squid project related to two denial of service security advisories which they say could affect all reverse proxies. http://www.squid-cache.org/Advisories/SQUID-2018_1.txt http://www.squid-cache.org/Advisories/SQUID-2018_2.txt For example frontier server launchpad squids are configured as reverse proxies. * Fri Dec 08 2017 Dave Dykstra <dwd@fnal.gov> 3.5.27-2.1 - Upgrade to 3.5.27-2 tarball with the following release note: - Add openhtc.io aliases to CMS_FRONTIER, ATLAS_FRONTIER, and MAJOR_CVMFS acls. * Fri Sep 01 2017 Dave Dykstra <dwd@fnal.gov> 3.5.27-1.1 - Upgrade to 3.5.27-1 tarball with the following release notes: - Upgrade to squid-3.5.27. The release announcement hasn't been published yet but here is the ChangeLog: https://github.com/squid-cache/squid/blob/v3.5/ChangeLog Release notes from prior releases since last frontier-squid release: https://www.spinics.net/lists/squid/msg86349.html (3.5.26) https://www.spinics.net/lists/squid/msg85634.html (3.5.25) - Remove the patch from #2833 because it was updated and included in the 3.5.27 release. - Add a patch from #4767 for making IPv6 SNMP queries work with multiple worker processes instead of causing a crash & restart. - Add IHEP and UNL stratum ones to the MAJOR_CVMFS access control list. * Wed Mar 22 2017 Dave Dykstra <dwd@fnal.gov> 3.5.24-3.1 - Upgrade to 3.5.24-3 tarball with the following release notes: - Change quick_abort_min and quick_abort_max once again, this time to 0 KB, because someone doing a different repeatable test found that the 1 GB value still caused some crashes but 0 KB didn't. * Fri Mar 17 2017 Dave Dykstra <dwd@fnal.gov> 3.5.24-2.1 - Upgrade to 3.5.24-2 tarball with the following release notes: - Change the default setting of quick_abort_min from -1 KB to 1 GB and the default setting of quick_abort_max from 16 KB to 1 GB. This is to work around the squid crash documented in squid bug #4554. A consequence is that when a client aborts an object download, the object won't get cached even though it is highly likely that another client will soon want the same object. * Tue Jan 31 2017 Dave Dykstra <dwd@fnal.gov> 3.5.24-1.1 - Upgrade to 3.5.24-1 tarball with the following release notes: - Upgrade to squid-3.5.24 with release notes at https://www.spinics.net/lists/squid/msg85024.html The bug fixes we had been including are not yet in this release, so all the same patches are kept. The most notable change affecting WLCG in this release was a fix for a bug that caused 'cache deny' to not have any effect. That is used for some cvmfs stratum 1s, and perhaps other applications. * Thu Jan 26 2017 Dave Dykstra <dwd@fnal.gov> 3.5.23-6.1 - Upgrade to 3.5.23-6 tarball with the following release notes: - Replace the bug #2833 fix with a newer version from the consultant. The full name is SQUID-252-collapsed-slaves-non-sharable-responses-3.5-t10. This one might be final. - Add patch from squid bug #4616, since two users had reported the "mem" assertion failure in store_client.cc. The full name of the patch is bug4616-cf-mem-assert-t1. * Wed Jan 25 2017 Dave Dykstra <dwd@fnal.gov> 3.5.23-5.1 - Upgrade to 3.5.23-5 tarball with the following release note: - Replace the bug #2833 fix with a newer version from the consultant. The full name is SQUID-252-collapsed-slaves-non-sharable-responses-3.5-t9. The previous version is suspected in a crash at one grid site. This version required a couple of small changes to apply and compile cleanly. * Fri Jan 13 2017 Dave Dykstra <dwd@fnal.gov> 3.5.23-4.1 - Upgrade to 3.5.23-4 tarball with the following release notes: - Replace the bug #2833 fix with a newer version from the consultant. The full name is SQUID-252-collapsed-slaves-non-sharable-responses-3.5-t6. It is probably not the final patch. - Add creation of /var/run/squid in the init.d script, if it did not exist. This is because systemd-based systems (EL7) do not preserve the directory across boots, and on some systems the systemd configuration file to create it at boot time does not work. That can happen if the system gets user ids from the network, so the squid user id might not yet be available when the systemd configuration file is used. - Add a systemd service descriptor configuration, primarily to make sure that frontier-squid is started after networking services. Uses the same "After" and "Wantedby" targets as apache httpd. * Tue Dec 27 2016 Dave Dykstra <dwd@fnal.gov> 3.5.23-3.1 - Upgrade to 3.5.23-3 tarball with the following release note: - Replace my patch that comments out the fix for security advisory SQUID-2016:10 with a patch from the consultant for bug #2833. This patch makes collapsed forwarding work again with If-Modified-Since while leaving the security advisory fixed. * Tue Dec 20 2016 Dave Dykstra <dwd@fnal.gov> 3.5.23-2.1 - Upgrade to 3.5.23-2 tarball with the following release note: - Comment out the fix for security advisory SQUID-2016:10, because it caused collapsed forwarding to break for If-Modified-Since requests. It was probably implemented incorrectly; discussion is in the reopened squid bug report #2833. * Mon Dec 19 2016 Dave Dykstra <dwd@fnal.gov> 3.5.23-1.1 - Upgrade to 3.5.23-1 tarball with the following release notes: - Upgrade to squid-3.5.23 with release notes at http://www.spinics.net/lists/squid/msg84536.html The most notable changes are fixes for two privacy vulnerabilities. They don't affect the applications primarily used with frontier-squid but may affect other applications. - Add the CERN IPv6 monitoring address range to the HOST_MONITOR acl. - Add IPv6 private net addresses to default setting of NET_LOCAL. * Fri Oct 14 2016 Dave Dykstra <dwd@fnal.gov> 3.5.22-2.1 - Upgrade to 3.5.22-2 tarball with the following release note: - Change default value of dns_v4_first to "on", so it will always try ipv4 first if available and then ipv6. The default was "off" which always tries ipv6 first. * Thu Oct 13 2016 Dave Dykstra <dwd@fnal.gov> 3.5.22-1.1 - Upgrade to 3.5.22-1 tarball with the following release notes: - Upgrade to squid-3.5.22 with release notes at http://www.spinics.net/lists/squid/msg83740.html - Remove patches from Alex Rousskov because they are now in the official squid 3.5 release. * Tue Sep 20 2016 Dave Dykstra <dwd@fnal.gov> 3.5.21-2.1 - Upgrade to 3.5.21-2 tarball with the following release note: - Fix error in the new logformat. It should be "%>Hs", not ">%Hs". * Tue Sep 13 2016 Dave Dykstra <dwd@fnal.gov> 3.5.21-1.1 - Upgrade to 3.5.21-1 tarball with the following release notes: - Upgrade to squid-3.5.21 with release notes at http://www.spinics.net/lists/squid/msg83169.html - Remove patch for bug #4428, it is now in the release. - Apply patch from Alex Rousskov for squid bug #4471, fixing collapsed forwarding revalidation when there is no If-Modified-Since. - Change "%Hs" in the default log format to the new style ">%Hs" for squid-3. - Change the default cache_dir size in squid.conf.proto to 10000 MB in case someone deletes the default 10000 MB line in customize.sh. - Change the script that applies patches to exit with an error if any of the patches are not completely applied. * Mon Aug 22 2016 Dave Dykstra <dwd@fnal.gov> 3.5.20-3.1 - Upgrade to 3.5.20-3 tarball with the following release notes: - Apply patch from the squid team for bug #4428. This bug causes the Cache-control: stale-if-error header, which Frontier uses, to be malformed in cached objects. - Fix bug that caused access logs shared by multiple workers to be rotated multiple times, once for each worker, instead of just once. - When using compressed logs and SQUID_CLEAN_CACHE_ON_START is true (both of which are default), then truncate the swap.state file in ufs cache directories each time logs are rotated. Otherwise the file grows without bounds. - When using the 'restart' function, clean ufs cache directories the same way as when doing 'start'. * Mon Aug 08 2016 Dave Dykstra <dwd@fnal.gov> 3.5.20-2.1 - Upgrade to 3.5.20-2 tarball with the following release notes: - The infamous squid bug #7 had partially cropped up again. The Date headers on cached 304 Not Modified responses on large objects was returning the original Date rather than a new re-validated Date, even though the Age header was correct. This causes many duplicated queries to get sent upstream. This release works around the problem as discussed here: http://bugs.squid-cache.org/show_bug.cgi?id=7#c79 * Thu Jul 21 2016 Dave Dykstra <dwd@fnal.gov> 3.5.20-1.1 - Upgrade to 3.5.20-1 tarball with the following release notes: - Upgrade to squid-3.5.20. This is the release announcement: http://www.spinics.net/lists/squid/msg82165.html - Replace my own patches for squid bugs #4311 and #4471 with a big patch from Alex Rousskov fixing collapsed revalidation. It addresses bug #4311 and more importantly bug #2833, where collapsed forwarding did not work with If-Modified-Since on expired objects. Unfortunately it breaks the functionality formerly fixed in the patch for bug #4471; that is, collapsed forwarding now does not work on expired objects that do *not* have a Last-Modified header. This should be no problem for the Frontier application because any response without Last-Modified is given a year expiration time. However, this is a potential issue for CVMFS because the stratum 1s currently expire most objects after 3 days. At least the issue will be relatively infrequent, and if a fix for this bug does not come soon we can probably eventually get the stratum 1 configurations changed to increase the expiration time. * Wed Jul 13 2016 Dave Dykstra <dwd@fnal.gov> 3.5.19-3.2 - Disable external acl helpers, auth digest helpers, and the LDAP auth basic helper, because they interfered with having an el5 rpm be upward compatible to el7 (in particular, libsasl and libdb). * Thu Jun 16 2016 Dave Dykstra <dwd@fnal.gov> 3.5.19-3.1 - Upgrade to 3.5.19-3 tarball with the following release note: - Change default minimum_expiry_time to 0 seconds. Without this change, squid will not cache any objects that are loaded for the first time during the last minute of their lifetime; instead, all requests for that object during that minute are sent upstream. It only affects squids that are fed by other squids. * Fri Jun 10 2016 Dave Dykstra <dwd@fnal.gov> 3.5.19-2.1 - Upgrade to 3.5.19-2 tarball with the following release note: - Add support for SQUID_CUSTOMIZE environment variable. If set to false it skips generating squid.conf from squid.conf.frontierdefault and customize.sh. Instead, the user is on his or her own to manage squid.conf. It is recommended to start from squid.conf.default because that is less likely to change between releases. The default setting of SQUID_CUSTOMIZE is true. * Mon May 23 2016 Dave Dykstra <dwd@fnal.gov> 3.5.19-1.2 - Add "Conflicts: frontier-awstats < 6.9-4" in order to ensure there's no version that does not support IPv6. * Wed May 11 2016 Dave Dykstra <dwd@fnal.gov> 3.5.19-1.1 - Upgrade to 3.5.19-1 tarball with the following release notes: - Upgrade to squid-3.5.19. This is the release announcement: http://www.spinics.net/lists/squid/msg81196.html - Add hepvm.cern.ch to the MAJOR_CVMFS acl * Mon Mar 28 2016 Dave Dykstra <dwd@fnal.gov> 3.5.15-2.1 - Add --disable-arch-native configure option to avoid illegal instruction errors as documented at http://wiki.squid-cache.org/KnowledgeBase/IllegalInstructionError - Upgrade to 3.5.15-2 tarball with the following release notes: - Disable memory_cache_shared by default, and make it an error if somebody tries to turn it on without rock cache. - Replace patch for bug #4312 with the one for #4311. It has the same effect and is simpler. * Thu Mar 24 2016 Dave Dykstra <dwd@fnal.gov> 3.5.15-1.1 - Upgrade to 3.5.15-1 tarball with the following release notes: - Upgrade to squid-3.5.15. This is the release announcement: http://www.spinics.net/lists/squid/msg79992.html - Apply patch for squid bug #2831, to send Cache-control header through on 304 Not Modified responses. - Apply patch for squid bug #4471, to make collapsed forwarding work when there's a previously cached but expired object. It still does not work with If-Modified-Since 304 Not Modified responses (bug #2833). So this makes it work with CVMFS but not yet Frontier. * Thu Oct 15 2015 Dave Dykstra <dwd@fnal.gov> 3.5.9-1.1 - Upgrade to 3.5.9-1 tarball with the following release notes: - Upgrade to squid-3.5.9. This is the release announcement: http://www.spinics.net/lists/squid/msg77242.html - Pass the SQUID variables in /etc/sysconfig/frontier-squid to the daily log rotation cron, in particular so SQUID_NUM_SERVICES will apply. - When installing with the "proto_install" make target, for rpm, install cron scripts in /usr/share/squid/cron instead of /etc/squid/cron. - Move scripts that were in /etc/squid/cron to /usr/share/squid/cron. - Removed the %config(noreplace) on /etc/cron.d/frontier-squid.cron, because that referred to the above path and it has to get changed. Also it was always difficult to allow administrator edits because the rpm also edits it when $FRONTIER_USER is set to something other than 'squid'. * Wed Sep 16 2015 Dave Dykstra <dwd@fnal.gov> 3.5.7-1.1 - Add creation of config file in /usr/lib/tmpfiles.d to create /run/squid directory on reboot of EL7-based systems. - Remove creation of /etc/squid/errors symlink, it's not needed by squid3. - Upgrade to 3.5.7-1 tarball with the following release notes: - Upgrade to squid-3.5.7. This is the release announcement: http://www.spinics.net/lists/squid/msg76566.html - Extend the "setoption" macro in customize.sh to work with options that are only identified by a "TAG:" comment in squid.conf, because not all options have a commented example like they did in squid2. - Add support for workers > 1, especially including recognizing the SMP macros that split up directories for separate worker processes. Ensure that the user has either enabled this for the cache directory or chosen rock cache type. For details on the SMP macros see "SMP-Related Macros" in squid.conf comments. - Enable support for the "rock" cache_dir store type, for sharing a cache between multiple SMP workers. Note that this cache type is still susceptible to squid bug #7 so it should be considered only experimental at this point. It may be safely used if it is guaranteed that no application will use If-Modified-Since (for example, if it is only used for CVMFS). - Enable support for the "diskd" cache_dir store type, which is just like "ufs" except that it uses a separate process for disk i/o. - Add support for running multiple independent squid services via the environment variable SQUID_NUM_SERVICES. This assigns each squid a "service name" that is a number from 0 to $SQUID_NUM_SERVICES-1. This value can be accessed with the macro ${service_name} which must be included in the cache_dir, access_log, cache_log, and pid_filename options. Add new customize.sh macro "setserviceoption" for setting options with a numerical value (e.g. http_port) to a different value per service. The value may also be a comma separated list of numbers (e.g. cpu_affinity_map). For usage see customhelps.awk. Each service may have multiple workers if desired. - Remove old wrapper script support for multiple processes since squid3 does everything natively in squid.conf. That includes removing support for automatically setting the core affinity on the separate processes. Instead set the option cpu_affinity_map to get the most performance out of each core. Setting core affinity improves performance by about 15%. - Reduce the verbose debug messages that were coming out at startup. - Add a patch for bug http://bugs.squid-cache.org/show_bug.cgi?id=3952 to prevent the initialization of the cache directory with multiple workers from running in the background. - Add a patch for bug http://bugs.squid-cache.org/show_bug.cgi?id=4312 to add a configuration option collapsed_forwarding_shared_entries_limit. This enables controlling the sharing of collapsed forwarding between SMP workers. This sharing causes deadlocks with the default ufs cache type, so this is set to zero in the default squid.conf. It does not cause deadlocks with rock cache, so to enable it use the comment() macro in customize.sh to drop back to the compiled-in default. - Extend the rotate lock to also cover awstats generation, to prevent more than one from running at a time. * Fri May 22 2015 Dave Dykstra <dwd@fnal.gov> 3.5.4-1.1 - Remove selection of PAM basic auth module at configure time because it could not be built. - Upgrade to frontier-squid-3.5.4-1 tarball with the following release notes: - Upgrade to squid-3.5.4, including a fix for a security vulnerability, release announcement here: http://www.spinics.net/lists/squid/msg75017.html - Merge changes from frontier-squid 2.7.STABLE9-22-1 and 2.7.STABLE9-23-1, which were (in addition to multiple-squid related things which don't function in frontier-squid3): - Support a "daemon:" prefix on access_log and cache_log, a poorly documented squid feature that uses a separate process to handle writing to log files so the main squid process doesn't have to wait for disk I/O. This was added because log compression was observed on one machine to interfere with squid I/O accesses. Make this the default for access_log. - Run log rotation with ionice -n7. - Make slight correction to MAJOR_CVMFS acl regular expression. - Expand the server names allowed at RAL in the ATLAS_FRONTIER acl. * Mon Apr 06 2015 Dave Dykstra <dwd@fnal.gov> 3.5.3-1.1 - Upgrade to frontier-squid-3.5.3-1 tarball with the following release notes: - Upgrade to squid-3.5.3, release announcement here: http://lists.squid-cache.org/pipermail/squid-announce/2015-March/000014.html * Thu Feb 19 2015 Dave Dykstra <dwd@fnal.gov> 3.5.2-1.1 - Upgrade to frontier-squid-3.5.2-1 tarball with the following release notes: - Upgrade to squid-3.5.2, release announcement here: http://lists.squid-cache.org/pipermail/squid-announce/2015-February/000012.html * Wed Jan 21 2015 Dave Dykstra <dwd@fnal.gov> 3.5.1-1.1 - Upgrade to frontier-squid-3.5.1-1 tarball with the following release note: - Upgrade to first full squid3.5 release, 3.5.1 * Mon Dec 22 2014 Dave Dykstra <dwd@fnal.gov> - Upgrade to frontier-squid- tarball which has the following release notes: - Upgrade to squid beta release, release announcement here: http://lists.squid-cache.org/pipermail/squid-announce/2014-December/000007.html - Merge changes from frontier-squid-2.7.STABLE9-21 which were: - Fix redirection of stderr in the hourly and daily crons so error messages properly go to squidcron.log. - Only require the minimum 4096 file descriptors when doing one of the commands that contain "start" (that is, "start", "restart", or "condrestart") with multiple squids. - Add the script that generates squid.conf to the list of files that trigger regenerating squid.conf if they're newer than squid.conf. - Only generate the per-squid configuration files used with multiple squids when squid.conf is newer than them. - NOTE: starting multiple squids the frontier-squid way (with squidN subdirectories under cache directory) doesn't work with squid-3, but so far the startup script capability is preserved just to keep as much as possible in common with frontier-squid-2 for now. Instead, to run multiple processes set the 'workers' configuration option. - Merge change from a pre-release of frontier-squid-2: - Make slight correction to MAJOR_CVMFS acl regular expression * Mon Nov 10 2014 Dave Dykstra <dwd@fnal.gov> - Upgrade to frontier-squid- tarball which has the following release note: - Upgrade to squid beta release Made as minimal changes as possible to the frontier-squid packaging for an early first look at a squid-3 with most of the features required by grid applications. * Thu Nov 06 2014 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-20.1 - Upgrade to frontier-squid-2.7.STABLE9-20 tarball which has the following release notes: - Increase the maximum number of squids that may be started from 4 to 16 - When running N > 1 squids, limit each squid to the hard limit on file descriptors divided by N. This limiting greatly reduces or eliminates the number of failed accesses to cache files indicated by TCP_SWAPFAIL_MISS entries in access.log. Require a minimum of 4096 file descriptors for each squid, unless customize.sh sets a value below the calculated limit. - Support use of SQUID_SUFFIX to add a suffix to all the files. This is not supported for use in the standalone tarball, just for use within an rpm. - Add logrotate as a dependency * Tue Sep 16 2014 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-19.1 - Upgrade to frontier-squid-2.7.STABLE9-19 tarball, which has the following release notes: - Make the default SQUID_MAX_ACCESS_LOG be 5G instead of 1G unless log compression is disabled. This should take about the same maximum space (~11Gbytes) as uncompressed log files did with a max size of 1G. - Add the TRIUMF CVMFS stratum 1 to the list in MAJOR_CVMFS. * Wed Sep 03 2014 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-18.1 - Upgrade to frontier-squid-2.7.STABLE9-18 tarball, which has the following release notes: - Protect rotate operations with a lock, because now that they compress files they can take a long time. This is especially important for the one minute between the daily cron and the first hourly cron. - Fix bug introduced in last release where if access_log is set to "none", the cache log is rotated every 15 minutes. * Fri Aug 22 2014 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-17.1 - Remove the cache when doing a full uninstall - When $FRONTIER_USER is not the default value 'squid', automatically manage the /etc/cron.d/frontier-squid.cron.rpmnew file that is always created - Upgrade to frontier-squid-2.7.STABLE9-17 tarball, which has the following release notes: - Update the CERN Hungary Data Center's LHCOPN IP address range in the HOST_MONITOR access control list from the incorrect to the correct and - Add commented-out acls CMS_FRONTIER, ATLAS_FRONTIER, and MAJOR_CVMFS that can be uncommented and used in place of RESTRICT_DEST to restrict outgoing connections to the corresponding servers. This allows updating the lists via frontier-squid package upgrades rather than requiring individual administrators to know how to keep the lists up to date. - Include the real time zone in the access.log timestamp instead of always +0000, and include milliseconds after the seconds. - Add the "cvmfs-info" header to the same double-quoted log entry that now has "X-Frontier-Id". Since no client sends both headers, only one will show at a time; frontier entries will end with " -" and cvmfs entries will start with "- ". Cvmfs clients currently only send cvmfs-info if configured with CVMFS_SEND_INFO_HEADER=yes so if that's not the case their log entries will show "- -". - Accept SQUID_MAX_ACCESS_LOG as the variable setting the maximum access log file size in place of LARGE_ACCESS_LOG (which is still accepted for backward compatibility). Also if the value ends in 'M' it indicates megabytes and if it ends in 'G' it indicates gigabytes; the default is bytes. - Run the "hourly" cron 4 times an hour, to catch faster when a log file has gone over the max size limit. - Compress log files by default, using logrotate. If environment variable SQUID_COMPRESS_LOGS is exported and set to 'false', fall back to the previous method of telling squid to rotate the log files without compression. In either case the logfile_rotate configuration parameter is used to define the maximum number of rotated files. If frontier-awstats is also installed, the first file is left uncompressed. When switching either way between compressed and uncompressed, removes all log files of the old type. - Rotate cache.log even if the access_log configuration parameter is "none". - Add a new "removecache" option to the init script to simply remove all of the cache, for use when removing the package. * Fri May 10 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-16.1 - Fix /etc/init.d/frontier-squid so if an alternate $FRONTIER_USER is set in /etc/squid/squidconf, then it will verify the existence of that user's home directory rather than the squid user's home directory. - Update to frontier-squid-2.7.STABLE9-16 tarball which rearranged the /etc/init.d script to be easier to modify at post-install time by the rpm, and which changed the delay after starting squid from 3 to 10 seconds before checking to see if it is running. That was the previous delay before the changes in the last release, and a system with 2 squids and slow disk access while cleaning cache in the background apparently didn't get started within 3 seconds. * Thu May 09 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-15.1 - Create the squid user's home directory if it doesn't exist at install time, because on RHEL6-derived systems a missing home directory prevents cron from running the squid log rotation. - Update to frontier-squid-2.7.STABLE9-15 tarball which has the following release notes: - Put squidcron.log in the same directory as cache.log rather than access.log, in case the access_log option is set to "none". In the previous version it would put squidcron.log in the squid user's home directory if access_log was "none". - Change the init.d startup script to abort with an error message if the squid user's home directory does not exist, because on RHEL6-derived systems if a user's home directory doesn't exist then cron won't run the user's jobs. - Run squid without the -S option so it will never run an audit of the cache files. During a normal start the cache is deleted so it doesn't matter, and the audit operation can take a very long time on a large cache during a restart. Also an analysis showed that the typical types of errors the audit catches (missing files) are survivable. - Allow multiple background cache cleans to be happening at the same time, in case the cache is very large and someone does multiple stop/start operations. - Add environment variable SQUID_CLEAN_CACHE_ON_START which defaults to true and when set to false prevents clearing the cache on start. It can be set and exported in the package's /etc/sysconfig file. * Wed Feb 06 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-13.5 - Change /var/cache/squid and /var/log/squid to be %ghost and created in the post install step in case someone had relocated them with a symlink. A symlink in one of those places with the previous version caused the target of the symlink to get changed to root ownership. * Mon Feb 04 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-13.4 - Add /var/cache/squid, /var/log/squid, and /var/run/squid to the list of directories installed; they had been accidentally left out * Mon Feb 04 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-13.3 - Change the chkconfig behavior to more precisely match Redhat squid's package; an administrator will now have to do "chkconfig frontier-squid on" instead of "chkconfig --add frontier-squid" - Do a few other internal cleanup things to frontier-squid.spec * Fri Feb 01 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-13.2 - Make /etc/squid/squid.conf a %ghost file so if someone is upgrading from a non-frontier squid it won't be overwritten * Fri Feb 01 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-13.1 - Update to frontier-squid-2.7.STABLE9-13 tarball which moves the output of cron jobs from daily.log in the directory the cron scripts are in (/etc/squid/cron) to the directory of the other squid logs (default /var/log/squid). - Remove the ability to use --prefix at rpm install time to relocate, because this greatly simplifies the rpm code including no longer needing to install .proto files with the package. - Change the cron jobs from being run from the squid user's /var/spool/cron (crontab) file to instead being run from /etc/cron.d/frontier-squid.cron. - When a user id (default "squid") needs to be created, create it as a system account (that is, with a UID lower than the value of UID_MIN in /etc/login.defs). - If the requested group (default "squid") is missing, create it just like a missing user id. - Set the ownership of all files under /etc/squid to be the squid user instead of leaving some as root. - Remove execute permission from files in /etc/squid that don't need it. - Remove bogus squid doc files from /usr/share/doc. - Move the restart of a running process from the post-uninstall to the post-install actions. Note: this has a side effect of causing it to restart twice the first time a package is upgraded from the previous style to this style. - Move the release notes to the rpm changelog. * Fri Jan 25 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-12.1 - Update to frontier-squid-2.7.STABLE9-12 tarball which only changes a comment in customize.sh so it can make more sense for the rpm - Change initial install to not start squid or do chkconfig -add to match what the standard Redhat squid rpm does - Change upgrades so squid will only be restarted if it was running at the time of the upgrade - Fix bug that prevented stopping squid when removing the rpm while squid is running * Mon Jan 14 2013 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-11.1 - Update to frontier-squid-2.7.STABLE9-11 tarball, which adds the Referer and User-Agent headers to the default logformat, and includes the upstream squid source so it self-contained. - Fix a rpm spec bug that kept rpmbuild -ba from working * Fri Dec 21 2012 Dave Dykstra <dwd@fnal.gov> 2.7.STABLE9-10.1 - Update to frontier-squid-2.7.STABLE9-10 tarball, which changes the default SNMP monitoring addresses accepted to be all of the WLCG addresses at the main CERN center and the backup CERN Hungary data center, and disables the ICP port by default - Remove sl5 from the rpm name, as the rpm works for both sl5 & sl6 * Wed Sep 05 2012 David Front <dfront@gmail.com> 2.7.STABLE9-8.1 - Change the init.d 'rotate' command to first remove the oldest log files, rather than asking squid to do it, because it can take a long time to delete large access logs and squid stops servicing requests during the rotate process. * Thu Aug 02 2012 David Front <dfront@gmail.com> 2.7.STABLE9-7.1 - Use a fix of Dave Dykstra: fixed a small bug in the frontier-squid tarball that prevented the hourly cron from reading /etc/sysconfig/frontier-squid for the setting of the LARGE_ACCESS_LOG variable, so it was impossible for the user to be able to control the size of the log files kept. * Tue Jun 26 2012 David Front <dfront@gmail.com> 2.7.STABLE9-6.1 - Use new release of frontier-squid tarball. The most significant new feature is that multiple squids can be started on the same port. For details of that feature and other tarball changes, see http://frontier.cern.ch/dist/frontier-squid-releasenotes.txt - Use more of the tarball directly, especially the proto files - Use the full version of the rpm to identify the squid process in SNMP monitoring - A small technical change at the spec file to support source rpms for frontier rpms - Use symbolic link to unproto.sh to simplify source tar ball creation * Wed Jun 13 2012 David Front <dfront@gmail.com> 2.7.STABLE9-5.24 - Following the recommendation of Dave Dykstra, change /etc/squid/squid.conf to be %ghost at the spec file, and hence not installed by the rpm but yes removed at rpm uninstall. As a result, the one line place holder of this file is no more needed, and also no need to check for small file when testing if this config file should be regenerated, at fn-local-squid.sh.proto. * Wed Apr 18 2012 David Front <dfront@gmail.com> 2.7.STABLE9-5.23 - Change the default NET_LOCAL at /etc/squid/customize.sh to be rather than Related: At release notes for frontier-squid-2.7.STABLE9-5: at http://frontier.cern.ch/dist/frontier-squid-releasenotes.txt: Remove the hidden default of always allowing incoming access to private network addresses Instead, make that default explicit in the first customize.sh that is generated so if the administrator doesn't want that it can be easily changed. * Wed Dec 21 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.22 - Change $AWSTATS_SCRIPT at fn-local-squid.sh.proto from /etc/awstats/squid_rotate_run_awstats.sh to /etc/awstats/run_awstats.sh * Tue Dec 20 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.21 - Undo the following, from frontier-squid-2.7.STABLE9-5.15, because not accepted by Dave Dykstra: 'Create squid.conf only if it does not exist or at start, not at stop or status.' Hence, once squid.conf has to be recreated, it will be created at each call to fn-local-squid.sh. A (unpleasent) consequence of this: If fontier-squid is running, and any of the following variables is going to be changed at the new configuration: 1 CACHE_DIR 2 CACHE_LOG 3 ACCESS_LOG 4 PID_FILENAME the new configuration that will be created at service frontier-squid stop will not be aware of the previou related value, and as a result may cause various failures. This rare situation should be handled manually by the administrator. The following help file may be consulted to do this: http://frontier.cern.ch/dist/rpms/frontierRpmInstallationError.txt * Sun Dec 18 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.20 - Conditionally call the awstats script, if it exists, to prevent related bugs and simplify the code * Sun Dec 11 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.19 - src/frontier-squid-utils/bin/fn-local-squid.sh.proto: Fix bug, change for $PIDINODE, as suggested by Dave Dykstra, at https://savannah.cern.ch/bugs/?89488#comment32 * Thu Dec 08 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.18 - Apply patch by Dave Dykstra on frontier-squid-utils/bin/fn-local-squid.sh.proto: https://savannah.cern.ch/support/download.php?file_id=22825 * Tue Nov 29 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.17 - Attempt to complete supporting rpm verify, command:'rpm -V': -- Do not remove *.unproto files (after usage). -- Do less attribute checks to files: /etc/squid/cachemgr.conf /etc/squid/customize.sh /etc/squid/squid.conf * Sun Nov 27 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.16 - Supporting rpm verify, command:'rpm -V': -- Add the prefix '%verify(not user group)' to each %files entry, in order to prevent complaints from 'rpm -V', like: .....UG. <fileName> -- Note, however, that 'rpm -V' may still create misleading complaints like: missing ... <fileName>.proto * Tue Nov 22 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.15 - Change default pid_filename to be /var/run/squid/squid.pid - Change directory and file defaults: ownership to be: frontierUser:frontierGroup, and mode to be: 755 - Attempt to correct bug: replace squid.conf.default by squid.conf.proto at src/frontier-squid-utils/etc - Refuse to answer a service status query while building a configuration file - Create squid.conf only if it does not exist or at start, not at stop or status - Fix bug at postinstall.sh, previously attempted to stop non existing service: squid * Thu Nov 10 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.14 - Fix bugs related to frontier group and automicatic creation of default log, run and cache dirs. - If /etc/squid/squidconf file is missing, do not create it but rather assume that the forntier user is squid * Thu Oct 27 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.13 - The group of frontier user is not controlled anymore. (Whatever group this user has, is used for chmod files from root to frontier user.) * Sun Oct 23 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.12 - Commenting out error if /etc/squid/squid.conf is writable * Fri Oct 21 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.11 - %defaultsquidgroup should be 'squid' and not 'users', as it was by mistake previously - Change the default of logs dir to be under standard /var/log rather than /var/logs as it used to be - Avoid directory names that begin by // * Thu Sep 22 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.10 - Do clean cache at initial restart * Tue Sep 20 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.9 - Attempt to handle an installation error related to a (mysterious) /etc/init.d/frontier-squid.sh sybmolic link by taking care to rm such files at rpm remove * Mon Sep 19 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.8 - Another code change attempting to avoid recursive erasal of / at rpm upgrade - rpm rearranged to use the files at frontier-squid tarball as much as possible, avoid duplicating such files where not needed - Fix failure to upgrade frontier-squid rpm while previous release is running, by replacing the command /sbin/service frontier-squid start by the command /sbin/service frontier-squid restart at postinstall.sh, in this case - In case of failure at %post, change the error messages to be more clear, referring to 'as if' installed, as explained at http://frontier.cern.ch/dist/rpms/frontierRpmInstallationError.txt * Fri Aug 12 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.7 - Attempt to avoid recursive erasal of / at rpm upgrade * Fri Jul 29 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.6 - Fix bug: Change order of post install operations, to prevent possible attempt to refer to file /etc/squid/squid.conf before it has been created. - Instead of removing /etc/squid/squid.conf.old at rpm uninstall, do: /bin/mv /etc/squid/squid.conf.old `mktemp /tmp/squid.conf.oldXXXX` * Wed Jul 27 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.5 - Fix bug at fn-local-squid.sh.template that prevents log rotation of squid logs if awstats rpm is not installed * Wed Jul 20 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.4 - For #3 at https://savannah.cern.ch/bugs/?77447: Like, squid standard rpm, do create the user with the safest permission (nologin as shell and no password). - Support non /data base dir (edited at /etc/squid/customize.sh) * Fri Jun 10 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.3 * Thu Apr 21 2011 David Front <dfront@gmail.com> 2.7.STABLE9-5.2 - Initial release Originating from http://grid-deployment.web.cern.ch/grid-deployment/flavia/frontier-squid-2.7.STABLE9-5.1.sl5.src.rpm Changes: - Added examples for customize.sh: /etc/squid/customize.sh.example_launchpad_CERN /etc/squid/customize.sh.example_non_launchpad_CERN