Package: xymon Version: 4.3.17-5 Severity: serious Tags: patch pending fixed-upstream
Before christmas, I've setup a new Jessie installation with Xymon. Since two or three weeks ago that installation is fed with all the monitoring data our current and Wheezy based production monitoring server gets. After approximately one week without restart, the xymond process on the Jessie installation grew to 15 GB of memory usage. Valgrind showed chunks of ca. 9 MB and 3.5 MB getting lost. Xymons typical activity interval is 5 minutes. (The 9 MB sound realistic: Summing up 9 MB every 5 minutes for a week would get 17.7 GB.) How quick the memory usage grows seems to depend on the installation and/or usage, too. I also have a 4.3.17-2 backport running on Wheezy for many months, and we only had a single case where it ran long enough (3 months without restart) so that the memory usage grew in areas where we noticed something. But that installation has way fewer hosts to monitor and hence fewer messages to handle (2 vs 28 incoming messages per second), so the memory leak didn't become that obvious as it is now. And the memory leak has been fixed upstream with their recent 4.3.18 release (03-Feb-2015). The 4.3.18 upstream changelog contains the following changelog entry: - xymond: Fix memory leak With cherry-picking the following two patches from upstream's 4.3.18 branch in SVN, the memory usage of the same installation stayed around 256 MB of RAM usage (fluctuating between 247 MB and 263 MB) after three hours despite it grew to over 300 MB after less than 30 minutes before. And Valgrind seems happy since then, too. Description:Fix memory leak in acknowledge handling. Cherry-picked from the 4.3.18 upstream release. Origin: http://sourceforge.net/p/xymon/code/7488/ Applied-Upstream: http://sourceforge.net/p/xymon/code/7488/ Reviewed-by: Axel Beckert <a...@debian.org> Index: xymon/xymond/xymond.c =================================================================== --- xymon.orig/xymond/xymond.c 2015-02-13 15:38:17.000000000 +0100 +++ xymon/xymond/xymond.c 2015-02-13 15:40:05.000000000 +0100 @@ -2365,6 +2365,8 @@ if (zombie->dismsg) xfree(zombie->dismsg); if (zombie->ackmsg) xfree(zombie->ackmsg); if (zombie->grouplist) xfree(zombie->grouplist); + if (zombie->lastchange) xfree(zombie->lastchange); + if (zombie->testflags) xfree(zombie->testflags); flush_acklist(zombie, 1); xfree(zombie); dbgprintf("<- free_log_t\n"); Description: Set an initial size for the xymondlog output Cherry-picked from the 4.3.18 upstream release. . Original commit message: "xymond: Set an initial size for the xymondlog output, to avoid realloc'ing the buffer too much. Fix a valgrind complaint. JC Cleaver" . Despite the harmless sounding commit message, this seems to fix a severe memory leak. Without this patch, valgrind reported the loss of about 12 MB every five minutes on a bigger installation. Origin: http://sourceforge.net/p/xymon/code/7489/ Applied-Upstream: http://sourceforge.net/p/xymon/code/7489/ Index: xymon/xymond/xymond.c =================================================================== --- xymon.orig/xymond/xymond.c 2015-02-13 16:50:42.000000000 +0100 +++ xymon/xymond/xymond.c 2015-02-13 17:21:42.000000000 +0100 @@ -3679,12 +3679,13 @@ } xfree(msg->buf); - logdata = generate_outbuf(NULL, logfields, h, log, acklevel); + logdata = newstrbuffer(20480); + logdata = generate_outbuf(&logdata, logfields, h, log, acklevel); addtobuffer(logdata, msg_data(log->message, 0)); msg->doingwhat = RESPONDING; msg->buflen = STRBUFLEN(logdata); - msg->bufp = grabstrbuffer(logdata); + msg->bufp = msg->buf = grabstrbuffer(logdata); } clear_filter(logfilter); I intend to upload 4.3.17-6 with these two patches, the added CVE-ID for the previous upload and one translation update. I will initially upload it to the DELAYED queue and will then request a freeze exception for Jessie. All of that is already checked in into the "jessie" branch at https://anonscm.debian.org/cgit/collab-maint/xymon.git/log/?h=jessie -- System Information: Debian Release: 8.0 APT prefers unstable APT policy: (990, 'unstable'), (600, 'testing'), (500, 'buildd-unstable'), (400, 'stable'), (110, 'experimental'), (1, 'buildd-experimental') Architecture: amd64 (x86_64) Foreign Architectures: i386 Kernel: Linux 3.18.0-trunk-amd64 (SMP w/1 CPU core) Locale: LANG=en_US.UTF-8, LC_CTYPE=en_US.UTF-8 (charmap=UTF-8) Shell: /bin/sh linked to /bin/dash Init: sysvinit (via /sbin/init) -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
