Conversation:
Notices
-
@verius @takeshitakenji CentOS tends to lag behind out of a desire to stick to "stable" versions, so it wouldn't surprise mme.
-
ncsd in particular seems to leak memory in large amounts and rather rapidly.
-
1.5GB after 10 minutes of operation is ridiculous.
-
I'd known it has leaks - it has for a while - so I was restarting it on the hourly crontab. But even after just 50min of operation it was easily eating up 2-4GB of RAM
Probably just going to disable it.
-
@takeshitakenji Just web browsers, usually, and things made by Oracle.
I wonder if this was written by Oracle.
-
What exactly is the point of caching user credentials anyways? Is the database being hit that often on a normal machine? Hell a webserver gets a lot more traction that way than most given things like FTP and it still is a very low use rate. Doesn't seem worth the tradeoff.
-
@verius @takeshitakenji Well, prior to taking ncsd down just now, I had 528MB free of 8GB allocated to that VM. Now, I have 6.08 GB free. Talk about a memory hog.
-
@verius Er, a proper bind setup would resolve this handily, with none of the memory leaks :)
-
@verius And in any case the hostname caching wasn't enabled to begin with here.
-
@verius I googled into this and there's a still open bug report for this behaviour in Debian from as far back as 2009. Apparently it has something to do with certain glibc environments and only occurs with certain combinations of installed libraries. Chalk another one up to the specific version dance problems Linux can have, I guess.