Dec 27, 2011:

11:15 AM Changeset [2100] by mitchb
Fix syslog-ng/systemd interaction, or "socket to me!" systemd creates and passes datagram sockets to daemons (at least by default). Fedora's default configuration for syslog-ng expected the socket passed in to be a stream socket (and we slurped that in forever ago). When syslog-ng sees that the socket it received isn't of the type it expected, instead of either coping or barfing, it just opens a new /dev/log socket of the type it wants, which is just hunky-dory for programs that connect after it disowns systemd's socket, but for anything that already connected to the datagram socket from systemd, they're logging to a useless socket that systemd isn't poll()ing, and thus can block (and, depending on what programs end up in this situation, potentially wedge your entire system). So, uh... let's not do that anymore. Fedora's also changing their default config. See RHBZ #742624

Dec 26, 2011:

4:15 AM Ticket #276 (Support postgres) created by adehnert
PostgreSQL is kinda neat. Also, if we supported it, Django South would …

Dec 19, 2011:

7:39 PM Changeset [2099] by achernya
Fix scripts-syslog-ng history, built but uncommitted changes
6:15 PM Changeset [2098] by achernya
Certificate renewal for debathena and impact

Dec 17, 2011:

4:50 AM Changeset [2097] by geofft
scripts-syslog-ng-config: Package r2095 and r2096
3:05 AM Changeset [2096] by geofft
d_zroot: Merge r2093 and r2094 into Debian copy
3:00 AM Changeset [2095] by geofft
d_zroot: Synchronize punted messages in Debian and Fedora copies
2:12 AM Changeset [2094] by geofft
d_zroot: make 'message redacted' message go to -c scripts-auto
12:00 AM Changeset [2093] by geofft
d_zroot: Don't accidentally let @recipients be empty d_zroot.pl fails to notice when the .k5login is nonexistent/empty, and when it runs ("zwrite", "-c", "scripts-spew", @k5login), it ends up giving zwrite no recipients, which causes the message to go to <scripts-spew,*,*>. This has the potential to leak sensitive logs on a misconfigured server. Fix this by redacting messages and also warning at startup if the .k5login is empty.

Dec 16, 2011:

10:20 PM Changeset [2092] by adehnert
Package Evan Broder's PyAFS
8:58 PM Changeset [2091] by adehnert
mock: enable the scripts repo while building

Dec 13, 2011:

12:17 AM Ticket #275 (cronload should error when run without arguments) created by adehnert
At the moment, if you run cronload with no arguments, it'll give you …

Dec 11, 2011:

7:31 AM Ticket #274 (OpenAFS systemd script doesn't have "AFS in use" check for restart) closed by geofft
fixed: Should be fixed in r2089, submitted upstream as …
4:57 AM Changeset [2090] by geofft
openafs.spec: Incorporate upstream fix to systemd-ized specfile Taken from commit 47b010d1e1d0e8bdbbcad73d5aedc19719d5d96a "redhat: make rpms build on pre-f15 again", aka http://gerrit.openafs.org/5813 .
4:33 AM Changeset [2089] by geofft
openafs.spec: Don't attempt to restart OpenAFS on upgrade (Trac: #274) systemd is actually rather capable of leaving the OpenAFS client in an incredibly broken state, thanks to its willingness to track services and kill their processes. We should not attempt to restart it on upgrade, whether a normal upgrade or a migration from SysV initscripts. In the former case, it's fine (and correct) for the old AFS to keep running; in the latter case, the unit file is capable of correctly shutting down an initscript-launched client. The same is true for the OpenAFS server. This brings the packaging in line with the SysV-style initscript code in the specfile, which does not attempt to restart the service, as well as with e.g. Debian's packaging, which uses --no-restart-on-upgrade.
12:41 AM Ticket #274 (OpenAFS systemd script doesn't have "AFS in use" check for restart) created by geofft
Every other initscript I used carefully avoids Thing Horrible if you …

Dec 7, 2011:

2:30 PM Changeset [2088] by ezyang
Vhost configuration for peoplesearch and bakerfoundation.
2:24 PM Changeset [2087] by ezyang
Certificates for peoplesearch and bakerfoundation (sans vhost configuration.)

Dec 5, 2011:

7:49 PM Changeset [2086] by ezyang
Postfix update.

Dec 2, 2011:

12:45 AM Changeset [2085] by andersk
Remove RSS <link> from sorry server page We got a report that some RSS readers (liferea) pick up http://scripts.mit.edu/rss/ from the <link> error messages, despite the HTTP 503 status code. The <link> has no business being here anyway.
Note: See TracTimeline for information about the timeline view.