FMC Update Failure, Not Enough var disk space

[jaewon]

I was upgrading from 6.4.0 to 6.6.0 of FMC.
The upgrade failed with a message as:

Total /var disk space needed for upgrade is 24853430
vardiskneeded has been adjusted to 12426715
Fatal error: Not enough var disk space available. You need at least 12426715K free to perform this upgrade. You have 5559680K free.
Exit return value = 1

So, I cleaned up the /var/log messages, and did 'upgrade_resume.sh', it shows a different error message:


We need 18942668 Kilobytes in /var/tmp. Available only 13631488. at /usr/local/sf/bin/install_update.pl line 625.
Removed MAKESELF script: /var/tmp/upgrade-patch/Cisco_Firepower_Mgmt_Center_Upgrade-6.6.0-90.sh


df -h shows:
Filesystem Size Used Avail Use% Mounted on
/dev/sda5 3.7G 1.8G 1.8G 50% /
none 16G 72K 16G 1% /dev
/dev/sda1 88M 43M 39M 53% /boot
/dev/sda7 238G 212G 15G 94% /var
none 16G 172K 16G 1% /dev/shm
tmpfs 16G 0 16G 0% /sys/fs/cgroup
tmpfs 16G 0 16G 0% /sys/fs/cgroup/pm


Have you guys seen this messages?

[Marvin Rhoads · Accepted solution]

It's commonly encountered due to several bugs. I usually look for the largest files using something like this (run as expert mode root user):

find /var -type f -exec du -Sh {} + | sort -rh | head -n 15

[jaewon]

This command is wonderful. I should have known this command. Thank you Marvin, I'll keep it.

We got a Cisco TAC, and he deleted very big files under the /var/sf/SRU --> Sourcefire Rule Updates
Then disk use of /var changed from 94% to 24%.

Then resuming by upgrade_resume.sh
another issue arosed,
[ForkPool] [2023-08-10 05:42:35] [11939] waiting for 1 processes to finish
DBI connect('DBN=vms;Host=127.0.0.1:10033;CharSet=utf8','DBA',...) failed: Database server not found (DBD: login failed) at /usr/local/sf/lib/perl/5.10.1/SF/SFDBI.pm line 375

then canceling this upgrade by using upgrade_rollback.sh

and then upgrade again, it worked.

I don't think all cases would be solved like this. Just for your information.