Resolved Upgrade from 18.0.34 to 18.0.35 causes container crash

[öndere]

Hello, when i try to upgrade my container from 18.0.34 to 18.0.35 cause my container crash / out of memory. Here is the error log.

[@bal ~]# find /tmp -type f -mtime +14 -exec rm {} \;
[@bal ~]# find /var/tmp -type f -mtime +14 -exec rm {} \;
[@bal ~]# df -i
Filesystem Inodes IUsed IFree IUse% Mounted on
/dev/root 3244032 171645 3072387 6% /
devtmpfs 250641 447 250194 1% /dev
tmpfs 251662 1 251661 1% /dev/shm
tmpfs 251662 651 251011 1% /run
tmpfs 251662 17 251645 1% /sys/fs/cgroup
tmpfs 251662 1 251661 1% /run/user/998
tmpfs 251662 1 251661 1% /run/user/0

 

[Nik G]

hello @öndere
as I can see there's enough inodes on your server, but log looks like there's no such disk space and due that fact the plesk database was corrupted during upgrade:

plesk-dovecot-2.3.13-2.centos.7+p18.0.34.2+t210325.1052.x86_64 Error in PREUN scriptlet in rpm package plesk-mail-pc-driver-18.0-2.centos.7+p18.0.34.2+t210325.1052.x86_64 Error in PREUN scriptlet in rpm package psa-mail-driver-common-18.0-2.centos.7+p18.0.34.2+t210325.1052.x86_64 error: Couldn't create temporary file for %preun(plesk-dovecot-2.3.13-2.centos.7+p18.0.34.2+t210325.1052.x86_64): No space left on device error: plesk-dovecot-2.3.13-2.centos.7+p18.0.34.2+t210325.1052.x86_64: erase failed error: Couldn't create temporary file for %preun(plesk-mail-pc-driver-18.0-2.centos.7+p18.0.34.2+t210325.1052.x86_64): No space left on device

[2021-04-30 18:33:43.953025] ERROR: Plesk\Exception\Database: DB query failed: SQLSTATE[HY000]: General error: 1021 Disk full (/var/tmp/#sql_456_0); waiting for someone to free some space..., query was: DESCRIBE `locales` (Mysql.php:79) [2021-04-30 18:33:43.947] ERR [panel] DB query failed: SQLSTATE[HY000]: General error: 1021 Disk full (/var/tmp/#sql_456_0); waiting for someone to free some space...

so you need to free space on your plesk server, restore psa DB from backup, so it become consistent before upgrade procedure and run upgrade process again.

 

[öndere]

Non-fatal POSTUN scriptlet failure in rpm package psa-mod_proxy-2.4.6-2.centos.7+p18.0.34.2+t210325.1052.x86_64
error: Couldn't create temporary file for %postun(psa-mod_proxy-2.4.6-2.centos.7+p18.0.34.2+t210325.1052.x86_64): No space left on device
Cleanup: plesk-dovecot-core [83/95]
Error in PREUN scriptlet in rpm package psa-mod_fcgid-2.3.9.5-2.centos.7+p18.0.34.2+t210325.1052.x86_64
error: Couldn't create temporary file for %preun(psa-mod_fcgid-2.3.9.5-2.centos.7+p18.0.34.2+t210325.1052.x86_64): No space left on device
Cleanup: plesk-modsecurity-crs [84/95]
error: psa-mod_fcgid-2error: psa-mod_fcgid-2.3.9.5-2.centos.7+p18.0.34.2+t210325.1052.x86_64: erase failed
error: Couldn't create temporary file for %preun(sw-nginx-1.18.0.2-2.centos.7+p18.0.34.2+t210325.1052.x86_64): No space left on device



Sooo strange. I have got enough space even if % 70 of container. Xa xa at the end result page showing none existing operating system (centos 7.9 .....)

 

[Nik G]

I have seen such situation inside virtuozzo containers when ploop was not used:

It can happen when somebody copied files from hardware node to the container's mounted root
in this case df utility said there's enough free space inside container.
but OS return an error that there's no space on device when you're trying to create file

I'm not talking that it is your case - it is just an example.

I guess you can start from verify that you really can utilize enough free space (for example create couple files in /var/tmp with total disk space used >= 4-5Gb)
in case successful I'd like to recommend you contact technical support of your hosting provider.

it looks like there are no problems from Plesk point of view, but something goes wrong on OS level.

 

[öndere]

I have seen such situation inside virtuozzo containers when ploop was no*t used:

It can happen when somebody copied files from hardware node to the container's mounted root
in this case df utility said there's enough free space inside container.
but OS return an error that there's no space on device when you're trying to create file

I'm not talking that it is your case - it is just an example.

I guess you can start from verify that you really can utilize enough free space (for example create couple files in /var/tmp with total disk space used >= 4-5Gb)
in case successful I'd like to recommend you contact technical support of your hosting provider.

it looks like there are no problems from Plesk point of view, but something goes wrong on OS level.

Ok thanks for your help. Issue is related with container's dns resolvers. Cloudflare dns blocked port 8447 so i changed it to linode's own dns resolvers. Fıxed