trouble pulling a remote repository.

Bill Barry after.fallout at gmail.com
Wed Jun 25 08:18:35 CDT 2008


Some further investigation reveals that the machine that created that 
revision fails hg verify the same way. That machine is running 
tortoisehg .4 rc2 and the rest of them are running rc1, but no hooks are 
being used and it is windows xp.

Matt Mackall wrote:
> On Tue, 2008-06-24 at 17:22 -0600, Bill Barry wrote:
>   
>> Matt Mackall wrote:
>>     
>>> You've shown us that at some point in the past you damaged your repo
>>> (which explains why your pull didn't work), but given us no hint how the
>>> original damage happened.
>>>
>>>   
>>>       
>> I don't know how the original damage happened.
>> This repo is only ever pushed to or pulled from and only over https
>> served by hgwebdir.cgi on apache.
>>
>> I suppose the following could have happened (in order of likelyness I
>> suppose):
>> 1. The network dropped a packet or two.
>> 2. Apache froze?
>> 3. The server failed to write a file?
>> 4. A tree fell in the forest and there was nobody around to hear it
>> (apparently it made a sound).
>>     
>
> Any of those should have left a partial transaction requiring 'hg
> recover'. In particular, the changelog is always written last when we
> push so there should never be a situation where we can see a changeset
> that hasn't been fully transferred.
>
> Anything interesting about your server setup? Hooks? Network filesystem?
> What OS is it on?
>
>   

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://selenic.com/pipermail/mercurial/attachments/20080625/8c4383d9/attachment.htm 


More information about the Mercurial mailing list