I too had a problem with syslog-ng 1.4.14 dying randomly on my RedHat Linux 7.2 box. I did a number of things and now it runs reliably. (at least for the last few weeks it hasn't had any problems) I don't know what fixed the problem but I will tell you what steps I took... I stoped using v1.4.14 and started using 1.5.17 In the config file options I added the following: time_reap(12); gc_busy_threshold(768); because in my particular environment I found that syslog-ng would have thousands of open files, this fixed that and lsof only shows hundreds of open files. I also started logging to a ReiserFS filesystem rather than ext2fs because I found that the directory scan performance on ext2fs slows to a crawl if you are constantly adding and removing files from a directory. (even when the directory was empty it was hundreds of KB) BTW, you might be interested in this: syslog-ng to SQL database HOWTO http://www.umialumni.com/~ben/SYSLOG-DOC.html -Ben.
On Wed, May 15, 2002 at 04:19:03PM -0400, Russo, Ben wrote: % I too had a problem with syslog-ng 1.4.14 dying randomly % on my RedHat Linux 7.2 box. [snip] % I stoped using v1.4.14 and started using 1.5.17 % In the config file options I added the following: % time_reap(12); % gc_busy_threshold(768); % because in my particular environment I found that % syslog-ng would have thousands of open files, this fixed % that and lsof only shows hundreds of open files. Tried these options; it still dies predictably when the syslog hub goes away. Thanks for the info, though. % I also started logging to a ReiserFS filesystem rather % than ext2fs because I found that the directory scan performance % on ext2fs slows to a crawl if you are constantly adding % and removing files from a directory. % (even when the directory was empty it was hundreds of KB) Yup; I'm doing the same for exactly that reason. john -- John Morrissey _o /\ ---- __o jwm@horde.net _-< \_ / \ ---- < \, www.horde.net/ __(_)/_(_)________/ \_______(_) /_(_)__
participants (2)
-
John Morrissey
-
Russo, Ben