0

I've got a server with real big hardware. (32 cores, 24Go Ram, MegaSAs Raid.... ).

It's running LAMP server, with memcached. It's running a fresh and up-to-date gentoo server, with custom kernel At some "random" times (I say random because that seems to not be link to loadAvg, peaks of disk/io or number of clients) the server seems to lag.

It can make lots of seconds to deliver simple assets files like js Mysql seems to freeze, with a big lot of threads, all stoped on "Sending Data".

In seconds, all could return to normal. Lots of monitoring, but nothing Obvious here. No network errors, D.N.S. config ok, no disk bottleneck ...

In this kind of situation, what could be good suspect ?

user1476033
  • 1

1 Answers1

0

CPU FAN (too little heating for 32 cores, so its slowing down), or SAS having some faults and recovering from them. Try to put it in power saving mode. Maybe you got dodgy harddrives, e.g. not SAS but some green desktop edition with power-saving feature.

Andrew
  • 1,037
  • 9
  • 17