But soon the very same problem started to happen to all of us.Finally we all got unable to doesn’t add much info we decide to attempt inspection of HTTP traffic to get more details. Fine, it grabs and depicts all the network communication - wow, it might come to dozens of packets in second for any usual computer.And even after filtering the relevant data with filters like it’s difficult to make anything from results. But - alas, - the tool shows only requests from browsers.
Mercurial updating to public failed
topic/httpfiddler/Im X0g6ufdl M: Here’s the first culprit: too long headers. Quite a large one, one would expect nicer things from Mercurial… After reviewing all those pages, the root cause becomes clear.
But after some thought it’s clear, that the client should send the server enough data for comparison between the two repositories, for server to send back only the missing changesets. Mercurial sees the repository as DAG of changesets taking into account the changeset itself, its parent(s) (generally one, two in case of merges) and children.
We’re nearing respectful number of 5000 commits which justifies the need for a larger request size. And here are some: https://bitbucket.org/site/master/issue/8263/http-400-bad-request-error-when-pulling, Branches, tags and other things come as a filtering aid.
I have a local copy of a Bitbucket repo on one of my servers (under the user rob), and I created a script that runs every hour in cron and attempts to pull & update the local copy.
The issue I'm having is that fails, even though when I follow the same procedure outside of cron it works fine.
I setup an ssh key for accessing Bitbucket in the normal fashion (i.e.After a couple of months of normal work with Mercurial repository served by IIS server, abruptly pulling attempts were responded with HTTP 400 error code.It took some time and skill to get the error details and discover the direct cause for this failure.Then we discovered that our branching workflow neglected some internal Mercurial logic. So I examined the build output briefly, understood it’s not for a couple of seconds and suggested to fix it tomorrow.Finally there are recommendations for a healthier use of branches. Coworker came and told me: “Team City fails; it can’t access repository. The error was like: And on the morrow I started investigating this.The problem didn’t reproduce on my computer, so I assumed it’s an internal Team City problem and tried some workarounds, like removing caches or recreating the repository or using different protocols.