Gitlab-CE Instance has poor performance

How to Use GitLab Self-managed
1

Hello,

We are using gitlab-ce 13.1.4 instance in our on premises infrastructure inside docker container. We have assigned it 8GB Memory & 6 CPU Cores.

We have integrated that gitlab with CI tools ( Jenkins & gitlab-ci ).

Problem: While Fetching code from gitlab repositories and pushing / pulling docker artifacts to / from gitlab instance we are facing very worst performance issues.

Also many times it’s throwing error like below.

Blockquote
fatal: The remote end hung up unexpectedly
fatal: early EOF
fatal: index-pack failed
Blockquote

Can Someone Please Assist here to add performance tuning ?

Regards,
Kushal Mehta

2

Do you have any graphs/numbers?

Without more to go on, my guess would be that 8GB of memory is (much) to little.

1 Like
3

Hello Grove,

Thanks for reply. Currently we are having our gitlab instance with few hundreds of git code repositories & many of them contains docker registry with docker container images ~1.4TB.
Would you suggest to do any special configuration inside gitlab.rb for puma worker ?

Regards,
Kushal

4

Have you made any changes to your network stack from the default config? Out of the box linux is net configured for high volume network traffic.

5

Hello,

No we have not changed any default configuration in any Kind of network stack. Also we are using centos as base OS. Please find below gitlab configuration for puma workers we are using , Let us know if we need to change.

puma[‘enable’] = true
puma[‘ha’] = false
puma[‘worker_timeout’] = 1200
puma[‘worker_processes’] = 2
puma[‘min_threads’] = 1
puma[‘max_threads’] = 4

6

Handling a 1.4TB image in just 8GB ram is going to be difficult. And you also need to handle other stuff in those 8GB.

We haven’t changed any settings regarding puma on our GitLab servers, and with only 8GB ram it’s not what I would focus on.

7

We have fixed this issue with Migrating docker registry to S3 bucket.
Thanks all for support.