On 4/29/2013 8:05 AM, James Harper wrote: >> On Mon, Apr 29, 2013 at 08:33:42AM +0000, James Harper wrote: >>> When loading the data into postgresql absolutely crawled along >> (~50kb/second disk write speed with 100% iowait) I knew I had a problem. >>> Something, somewhere has changed in my system that absolutely kills tiny >> sync writes. Or alternatively, something has changed in my system that >> makes mysql do tiny sync writes. >> >> What do you expect from sync writes regarding bacula? >> I don't use sync writes there as I am very sure they won't give me a >> benefit. As soon as the DB-server dies, the job is failed, you can't add >> the remaining attributes in a reasonable way later and sd/dir both >> refuse to work without a database server. >> I suggest turning sync writes off. >> This is valid for bacula, not for some random other database >> application. >> > The thing that bothers me is that it was working perfectly well up until a > week ago when it suddenly slowed to a crawl, and I can't figure out what > changed.
What happens if you run the same dd command without oflags=sync? Have you run e2fsck to see if there are any fs problems? How much free space is on that drive? Are there any kernel messages related to that drive? Is there a lack of free memory? It looks like a fs or hardware issue, since dd from /dev/zero won't even work. ------------------------------------------------------------------------------ Try New Relic Now & We'll Send You this Cool Shirt New Relic is the only SaaS-based application performance monitoring service that delivers powerful full stack analytics. Optimize and monitor your browser, app, & servers with just a few lines of code. Try New Relic and get this awesome Nerd Life shirt! http://p.sf.net/sfu/newrelic_d2d_apr _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
