Closed Bug 1499204 Opened 6 years ago Closed 6 years ago

Set ClientAliveInterval on SSH server

Categories

(Developer Services :: Mercurial: hg.mozilla.org, defect, P1)

defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: gps, Assigned: gps)

References

Details

Attachments

(1 file)

When aki attempted to push mozilla-beta to hg.mozilla.org today, SSH connections kept getting dropped while waiting for hooks to complete.

I think I successfully fingered the dropping of idle SSH connections as the culprit. (SSH will drop connections if no data has been exchanged for a while.)

On clients, you can enable ServerAliveInterval to send data every N seconds to keep the connection alive. Servers can enable ClientAliveInterval.

I hacked up the sshd_config for the hg server to add the following:

ClientAliveInterval 10
ClientAliveCountMax 6

aki's next push didn't drop.

We should make this change permanent so the server ensures idle SSH connections don't drop.

The settings above will drop the connection if the client doesn't respond to the pings after 60s. This seems reasonable. But I kinda pulled that number out of thin air. Not sure if others have a strong opinion on a better value...
Assignee: nobody → gps
Status: NEW → ASSIGNED
Flags: needinfo?(gps)
Priority: -- → P1
Blocks: 1489406
This will prevent clients from timing out due to no output from
server.
Pushed by cosheehan@mozilla.com:
https://hg.mozilla.org/hgcustom/version-control-tools/rev/c5cabbbc31a5
ansible/hg-ssh-server: enable ClientAliveInterval ; r=sheehan
Status: ASSIGNED → RESOLVED
Closed: 6 years ago
Resolution: --- → FIXED
Flags: needinfo?(gps)
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: