636103 - hg.m.o failure not handled correctly during tagging

Status: RESOLVED DUPLICATE of bug 575400

(Release Engineering :: General, defect)

Description

[Nick Thomas [:nthomas] (UTC+12)]

Attachment: Full tagging log

While tagging ku for Fx 4.0b12 build1 there was a problem with ssh connections to hg.m.o. The series of steps was basically
 # pull and tag successfully
 hg -q out ... http://hg.mozilla.org/l10n-central/ku   # successful

 hg -q out ... ssh://hg.mozilla.org/l10n-central/ku
 abort: no suitable response from remote hg!

 hg pull ssh://hg.mozilla.org/l10n-central/ku
 remote: ssh_exchange_identification: Connection closed by remote host^M
 abort: no suitable response from remote hg!

 # strip out two tagging changesets

So, if my very limited knowledge of the new tagging scripts still applies, we've treated the hg error on the hg out as a race, tried to pull in new changes and got another error, checked hg out again, and bailed out. I don't know if we can do this in a reliable way or not, but a retry at the hg out would have avoided the intermittent fault on hg.m.o. Or simply some fine-grained error handling on hg operations is needed.

Kudos for one locale not causing the rest of the repos to barf.

Comment 1

[[e-mail] (:bhearsum)]

bug 575400 is adding more retrying. Specifically, on all hg operations that touch the network, as well as on the entire tagging per-repository. Duping this to that.