Closed kondalarao15 closed 5 years ago
There must be something in your carbon log or system logs or coredump, etc.
Under carbon-cache-b/query.log . During that time carbin-cache-b process went down and I didn't see any exceptions in creates.log
Nothing in your system logs, no core dumps? Only time I've seen carbon go down with nothing in it's log was due to the Linux OOM killer.
I don’t have any other system logs, if you see in the first image for carbon process(it’s showing a to h) process but abruptly few of the carbon processes are going down. Can we have a quick meeting regarding this issue.
I don’t have any other system logs
What do you mean?
I've checked the system logs. These are the logs that are created on 05/12/2019 under /var/log folder.
Under messages I've seen one error
And
yup, oom-killer as expected. You are running out of memory, then Linux decides to kill a process.
That will do it, you're trying to run 8 caches and 2 relays so you probably don't have enough memory for all that on your host.
Currently I’m using 8 core machine with 16gb ram. Could you please suggest me what’s the ideal cpu and RAM requirement for running 8carbon cache and 2 relays.
Or else with current 8core machine with 16gb RAM, how many carbon caches I can run to avoid carbon downtime. On Wed, May 15, 2019 at 3:46 PM Dan Cech notifications@github.com wrote:
That will do it, you're trying to run 8 caches and 2 relays so you probably don't have enough memory for all that on your host.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/graphite-project/carbon/issues/854?email_source=notifications&email_token=ADJ2UND5GXPETIR27N3SUNLPVRZCVA5CNFSM4HNDXSB2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODVP4WYA#issuecomment-492817248, or mute the thread https://github.com/notifications/unsubscribe-auth/ADJ2UNCQIKKCU4AGUZTZBJDPVRZCVANCNFSM4HNDXSBQ .
Thanks piotr and DenCech. Currently I've carbon-cache-a to carbon-cache-h and two relays. Now I've commented carbon-cache-c to carbon-cache-h and restarted the graphite carbon. For some reason it's showing three carbon-cache-processes.
Please see the following file for carbon.conf configuration file carbon.txt
Could you please suggest me what’s the ideal cpu and RAM requirement for running 8carbon cache and 2 relays.
There are just too many variables to give an simple answer. Disk performance, number of metrics, metric interval, storage schema's, etc.
I've recently installed graphite and grafana on same instance and for some reason carbon processes going down with no error. I'm installed python 2.7.5 and graphite 1.1.5 versions.
I'm using the following carbon.conf file. carbon.txt
Current carbon process is up and running
But intermittently few of the carbon processes are going down and we're missing the metrics.