Log files, partitions and logrotate
f.cuseo
50 Posts
April 17, 2024, 8:12 amQuote from f.cuseo on April 17, 2024, 8:12 amHello.
I am using petasan mostly for S3 rados gateway.
The standard petasan partition scheme, reserve not a lot of space for / partition (/dev/sda3, 15G), and logs are in this partition.
With rados gateway, using veeam object storage repository, in less than 2 days the partition is 100% full, because of client.ceph.log and nginx logs.
I have more space in /dev/sda5 (/opt/petasan/config, 90G).
Looking at /etc/logrotate.d/ceph-common , I see that there is a 7 days rotate retention; is safe to move logfiles to /opt/petasan/config/ceph-logs/, create a symlink in /var/log/ceph, and reduce rotate to 1 day ? Or is better to use the "olddir" configuration to move in /opt/petasan/config/old-ceph-logs/ only old files ?
Looking at /etc/logrotate.d/nginx, i see 14 rotate days and a "delaycompress" statement (to have the last file uncompressed); will you use this file for statistics or something else or I can safely remove this statement, reduce rotate to 1 or 2 days, and add "olddir" configuration too ?
Will you change something in standard setup with 3.3 release ?
Thank you.
Hello.
I am using petasan mostly for S3 rados gateway.
The standard petasan partition scheme, reserve not a lot of space for / partition (/dev/sda3, 15G), and logs are in this partition.
With rados gateway, using veeam object storage repository, in less than 2 days the partition is 100% full, because of client.ceph.log and nginx logs.
I have more space in /dev/sda5 (/opt/petasan/config, 90G).
Looking at /etc/logrotate.d/ceph-common , I see that there is a 7 days rotate retention; is safe to move logfiles to /opt/petasan/config/ceph-logs/, create a symlink in /var/log/ceph, and reduce rotate to 1 day ? Or is better to use the "olddir" configuration to move in /opt/petasan/config/old-ceph-logs/ only old files ?
Looking at /etc/logrotate.d/nginx, i see 14 rotate days and a "delaycompress" statement (to have the last file uncompressed); will you use this file for statistics or something else or I can safely remove this statement, reduce rotate to 1 or 2 days, and add "olddir" configuration too ?
Will you change something in standard setup with 3.3 release ?
Thank you.
admin
2,858 Posts
April 17, 2024, 10:46 amQuote from admin on April 17, 2024, 10:46 amYes you could create sym link of /var/log to the large partition. changing logrotate will also be good. You can also disable rgw logging.
PetaSAN 3.3 will combine both partitions into 1, so it will not have this issue, but it is for new installs only.
Yes you could create sym link of /var/log to the large partition. changing logrotate will also be good. You can also disable rgw logging.
PetaSAN 3.3 will combine both partitions into 1, so it will not have this issue, but it is for new installs only.
Log files, partitions and logrotate
f.cuseo
50 Posts
Quote from f.cuseo on April 17, 2024, 8:12 amHello.
I am using petasan mostly for S3 rados gateway.
The standard petasan partition scheme, reserve not a lot of space for / partition (/dev/sda3, 15G), and logs are in this partition.
With rados gateway, using veeam object storage repository, in less than 2 days the partition is 100% full, because of client.ceph.log and nginx logs.
I have more space in /dev/sda5 (/opt/petasan/config, 90G).
Looking at /etc/logrotate.d/ceph-common , I see that there is a 7 days rotate retention; is safe to move logfiles to /opt/petasan/config/ceph-logs/, create a symlink in /var/log/ceph, and reduce rotate to 1 day ? Or is better to use the "olddir" configuration to move in /opt/petasan/config/old-ceph-logs/ only old files ?
Looking at /etc/logrotate.d/nginx, i see 14 rotate days and a "delaycompress" statement (to have the last file uncompressed); will you use this file for statistics or something else or I can safely remove this statement, reduce rotate to 1 or 2 days, and add "olddir" configuration too ?
Will you change something in standard setup with 3.3 release ?
Thank you.
Hello.
I am using petasan mostly for S3 rados gateway.
The standard petasan partition scheme, reserve not a lot of space for / partition (/dev/sda3, 15G), and logs are in this partition.
With rados gateway, using veeam object storage repository, in less than 2 days the partition is 100% full, because of client.ceph.log and nginx logs.
I have more space in /dev/sda5 (/opt/petasan/config, 90G).
Looking at /etc/logrotate.d/ceph-common , I see that there is a 7 days rotate retention; is safe to move logfiles to /opt/petasan/config/ceph-logs/, create a symlink in /var/log/ceph, and reduce rotate to 1 day ? Or is better to use the "olddir" configuration to move in /opt/petasan/config/old-ceph-logs/ only old files ?
Looking at /etc/logrotate.d/nginx, i see 14 rotate days and a "delaycompress" statement (to have the last file uncompressed); will you use this file for statistics or something else or I can safely remove this statement, reduce rotate to 1 or 2 days, and add "olddir" configuration too ?
Will you change something in standard setup with 3.3 release ?
Thank you.
admin
2,858 Posts
Quote from admin on April 17, 2024, 10:46 amYes you could create sym link of /var/log to the large partition. changing logrotate will also be good. You can also disable rgw logging.
PetaSAN 3.3 will combine both partitions into 1, so it will not have this issue, but it is for new installs only.
Yes you could create sym link of /var/log to the large partition. changing logrotate will also be good. You can also disable rgw logging.
PetaSAN 3.3 will combine both partitions into 1, so it will not have this issue, but it is for new installs only.