"Failed to write, and not due to blocking"
62 views
Skip to first unread message
Henrik Schröder
Aug 19, 2008, 5:52:18 AM8/19/08
to memc...@googlegroups.com
Hi everyone,
A short while ago, one of our memcached servers started dying, and if I run it in verbose mode (-vv) it prints "Failed to write, and not due to blocking" before exiting. I can find the place in the source where it does this, but the error message tells me very little, and I'm not familiar with the memcached code, could someone else fill me in on what exactly goes wrong? Some help on how to avoid the error would be nice as well... :-)
/Henrik
A short while ago, one of our memcached servers started dying, and if I run it in verbose mode (-vv) it prints "Failed to write, and not due to blocking" before exiting. I can find the place in the source where it does this, but the error message tells me very little, and I'm not familiar with the memcached code, could someone else fill me in on what exactly goes wrong? Some help on how to avoid the error would be nice as well... :-)
/Henrik
Henrik Schröder
Aug 19, 2008, 7:25:41 AM8/19/08
to memc...@googlegroups.com
Yeah, uh, I'm running version 1.2.5 on Windows, that's probably useful to know, I should not have forgotten that in my original email. :-)
I'm thinking that we get the error because of a recent change in our client where we do "gets" everywhere, and we do have some massive multi-gets, and retrieveing hundreds of values along with their CAS unique in a single get might cause some resource starvation.
I'm fixing our client now to properly use "gets" only when needed, and we'll see if that helps.
/Henrik
I'm thinking that we get the error because of a recent change in our client where we do "gets" everywhere, and we do have some massive multi-gets, and retrieveing hundreds of values along with their CAS unique in a single get might cause some resource starvation.
I'm fixing our client now to properly use "gets" only when needed, and we'll see if that helps.
/Henrik
Simone Busoli
Aug 24, 2008, 6:53:19 AM8/24/08
to memc...@googlegroups.com
Fixed?
Henrik Schröder
Aug 24, 2008, 8:46:19 AM8/24/08
to memc...@googlegroups.com
Yes, I did the changes in our client, and it works fine again now.
I'll see if I can make a test for memcached that consistently triggers this error, it seems like it happens when you do a "gets" on a lot of keys (100+). That ought to work just as well as doing a "get" on those same keys.
/Henrik
I'll see if I can make a test for memcached that consistently triggers this error, it seems like it happens when you do a "gets" on a lot of keys (100+). That ought to work just as well as doing a "get" on those same keys.
/Henrik
dormando
Aug 24, 2008, 2:45:17 PM8/24/08
to memc...@googlegroups.com
Can you look at the change history from 1.2.5 -> 1.2.6 and apply the CAS
crash fix? Maybe the rest of the crash fixes too? I don't know how hard
it'll be to update the windows port to 1.2.6, but someone really ought
to...
crash fix? Maybe the rest of the crash fixes too? I don't know how hard
it'll be to update the windows port to 1.2.6, but someone really ought
to...
-Dormando
Henrik Schröder
Sep 8, 2008, 12:32:27 PM9/8/08
to memc...@googlegroups.com
I tried to update the 1.2.5 Windows source I have with all the changes that went into 1.2.6, but it didn't compile and I'm completely unfamiliar with C programming on Windows so I have no idea why it didn't work, and not the skill to make it work. It would be awesome with an active windows maintainer and to get all the Windows changes into the main version, but I'm clearly not that person. Sorry. :-)
/Henrik Schröder
/Henrik Schröder
Reply all
Reply to author
Forward
0 new messages