User Tools

Site Tools


[TROUBLESHOOT] Ceph too many pgs per osd

Name: [TROUBLESHOOT] Ceph too many pgs per osd
Description: how to solve this “issue”
Modification date :11/04/2019
Notify changes to:Owner
Tags:ceph, object storage
Scalate to:The_fucking_bofh


This documents cover the TOO MANY PGs per OSD, not too few (documented here)

Understanding the problem

The solution is perfectly explained here:
To keep the solution safe, I'll write here down it.

Rawpaste of pool usage

ceph pg dump | awk '
 /^PG_STAT/ { col=1; while($col!="UP") {col++}; col++ }
 /^[0-9a-f]+\.[0-9a-f]+/ { match($0,/^[0-9a-f]+/); pool=substr($0, RSTART, RLENGTH); poollist[pool]=0;
 up=$col; i=0; RSTART=0; RLENGTH=0; delete osds; while(match(up,/[0-9]+/)>0) { osds[++i]=substr(up,RSTART,RLENGTH); up = substr(up, RSTART+RLENGTH) }
 for(i in osds) {array[osds[i],pool]++; osdlist[osds[i]];}
 printf("pool :\t"); for (i in poollist) printf("%s\t",i); printf("| SUM \n");
 for (i in poollist) printf("--------"); printf("----------------\n");
 for (i in osdlist) { printf("osd.%i\t", i); sum=0;
   for (j in poollist) { printf("%i\t", array[i,j]); sum+=array[i,j]; sumpool[j]+=array[i,j] }; printf("| %i\n",sum) }
 for (i in poollist) printf("--------"); printf("----------------\n");
 printf("SUM :\t"); for (i in poollist) printf("%s\t",sumpool[i]); printf("|\n");

That's all

Procedure for dropping pool: [HOWTO] Completely remove a POOL from cluster Don't drop the pool, continue reading.

The "solution"

If you're sure that you need more PG's per os, change the configuration.
You must add:

mon pg warn max per osd = ${NEW_PG_PER_OSD_NUMBER}

on your ceph.conf under the [global] section.

And update the config worldwide.

You can also change the setting without having to restart:

# check
ceph tell 'osd.*' config get mon_max_pg_per_osd
# change
ceph tell 'osd.*' config set mon_max_pg_per_osd ${NEW_PG_PER_OSD_NUMBER}
# check again
ceph tell 'osd.*' config get mon_max_pg_per_osd
linux/ceph/troubleshooting/too_many_pgs_per_osd.txt · Last modified: 2022/02/11 11:36 by