Segfault on CentOS about carbon-c-relay HOT 7 CLOSED

Hmmm, a backtrace would certainly help here, if possible.

What is high load for you? How does your config look like?

from carbon-c-relay.

The relay process occasionally spiked to ~400%CPU. Config is as follows:

cluster graphite3
fnv1a_ch
127.0.0.1:2103
127.0.0.1:2203
127.0.0.1:2303
127.0.0.1:2403
127.0.0.1:2503
127.0.0.1:2603
127.0.0.1:2703
127.0.0.1:2803
127.0.0.1:2903
127.0.0.1:3003
127.0.0.1:3103
127.0.0.1:3203
127.0.0.1:3303
127.0.0.1:3403
127.0.0.1:3503
127.0.0.1:3603 ;

cluster graphite2
fnv1a_ch
192.168.200.34:2103
192.168.200.34:2203
192.168.200.34:2303
192.168.200.34:2403
192.168.200.34:2503
192.168.200.34:2603
192.168.200.34:2703
192.168.200.34:2803
192.168.200.34:2903
192.168.200.34:3003
192.168.200.34:3103
192.168.200.34:3203
192.168.200.34:3303
192.168.200.34:3403
192.168.200.34:3503
192.168.200.34:3603 ;

match *
send to graphite3 ;
match *
send to graphite2 ;

How can I get a backtrace for you? Also, I'm not sure if I can currently duplicate the problem because we eliminated a lot of inbound metrics, so load has backed off quite a bit since then and the process has not segfaulted since.

from carbon-c-relay.

for a backtrace, compile with:
make CFLAGS="-g -pipe"
Then run manually in gdb:
gdb --args ./relay -f yourconf
when it crashes type
thread apply all bt

from carbon-c-relay.

Can you tell me what version you use(d)? If this is about a git version (e.g. not a release) it may not be a surprise at all. Yet the backtrace would still be useful.

from carbon-c-relay.

carbon-c-relay v0.36 (46d3a9)

I haven't been able to get it to segfault since I made some changes to the machine (improved disk throughput via RAID5, eliminated some unnecessary carbon-cache settings, etc.)

from carbon-c-relay.

I have no idea on how to fix this. The only thing I have left to ask you is you mention 400% cpu usage, on what system was that? How many CPUs did it have? Was there plenty of memory available?

from carbon-c-relay.

It's a single machine with 24 cores and 96GB RAM. It hasn't segfaulted since 1/27. I'm not able to provide any additional useful information on it, so I'll go ahead and close this issue.

from carbon-c-relay.