Is git://sourceware.org/git/binutils.git down?

Is git://sourceware.org/git/binutils.git down?

[H.J. Lu]

I got

[hjl@gnu-tools-1 binutils-linux]$ git fetch
fatal: read error: Connection reset by peer
[hjl@gnu-tools-1 binutils-linux]$


-- 
H.J.

Re: Is git://sourceware.org/git/binutils.git down?

[Frank Ch. Eigler]

Hi -

> [hjl@gnu-tools-1 binutils-linux]$ git fetch
> fatal: read error: Connection reset by peer
> [hjl@gnu-tools-1 binutils-linux]$

Use the ssh:// git URL to avoid the overload-shedding logic associated
with git://.

- FChE

Re: Is git://sourceware.org/git/binutils.git down?

[H.J. Lu]

On Tue, Nov 20, 2012 at 7:39 AM, Frank Ch. Eigler <fche@redhat.com> wrote:
> Hi -
>
>> [hjl@gnu-tools-1 binutils-linux]$ git fetch
>> fatal: read error: Connection reset by peer
>> [hjl@gnu-tools-1 binutils-linux]$
>
> Use the ssh:// git URL to avoid the overload-shedding logic associated
> with git://.
>

Something is wrong.  binutils git mirror hasn't been updated
for several hours and there were 5 commits missing in
git mirror.

-- 
H.J.

Re: Is git://sourceware.org/git/binutils.git down?

[Frank Ch. Eigler]

Hi -

On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
> Something is wrong.  binutils git mirror hasn't been updated
> for several hours and there were 5 commits missing in
> git mirror.

The way these mirrors are updated is by a rather CPU-expensive set of
scripts run by Jim Meyering.  I'm not too surprised if they sometimes
fall behind.

- FChE

Re: Is git://sourceware.org/git/binutils.git down?

[H.J. Lu]

On Tue, Nov 20, 2012 at 7:54 AM, Frank Ch. Eigler <fche@redhat.com> wrote:
> Hi -
>
> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>> Something is wrong.  binutils git mirror hasn't been updated
>> for several hours and there were 5 commits missing in
>> git mirror.
>
> The way these mirrors are updated is by a rather CPU-expensive set of
> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
> fall behind.
>

6 hours behind is very strange.


-- 
H.J.

Re: Is git://sourceware.org/git/binutils.git down?

[H.J. Lu]

On Tue, Nov 20, 2012 at 8:25 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
> On Tue, Nov 20, 2012 at 7:54 AM, Frank Ch. Eigler <fche@redhat.com> wrote:
>> Hi -
>>
>> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>>> Something is wrong.  binutils git mirror hasn't been updated
>>> for several hours and there were 5 commits missing in
>>> git mirror.
>>
>> The way these mirrors are updated is by a rather CPU-expensive set of
>> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
>> fall behind.
>>
>
> 6 hours behind is very strange.
>

Both gdb and binutils git mirrors are out of date.  Can someone
please take a look?

Thanks.

-- 
H.J.

Re: Is git://sourceware.org/git/binutils.git down?

[Jim Meyering]

H.J. Lu wrote:

> On Tue, Nov 20, 2012 at 8:25 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>> On Tue, Nov 20, 2012 at 7:54 AM, Frank Ch. Eigler <fche@redhat.com> wrote:
>>> Hi -
>>>
>>> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>>>> Something is wrong.  binutils git mirror hasn't been updated
>>>> for several hours and there were 5 commits missing in
>>>> git mirror.
>>>
>>> The way these mirrors are updated is by a rather CPU-expensive set of
>>> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
>>> fall behind.
>>>
>>
>> 6 hours behind is very strange.
>>
>
> Both gdb and binutils git mirrors are out of date.  Can someone
> please take a look?

There was one zombie mirror-sw process for each of those two
repositories, and due to the new locking mechanism[*], no other job
would proceed.  I've just killed all of the dependent processes
and that let the zombies die for real, so now things seem to
be moving again.

[*] This is the locking mechanism that Frank requested to
keep two update processes (for the same repo) from overlapping,
which was harmless, except for the load it imposed.

Re: Is git://sourceware.org/git/binutils.git down?

[H.J. Lu]

On Tue, Nov 20, 2012 at 8:19 PM, Jim Meyering <jim@meyering.net> wrote:
> H.J. Lu wrote:
>
>> On Tue, Nov 20, 2012 at 8:25 AM, H.J. Lu <hjl.tools@gmail.com> wrote:
>>> On Tue, Nov 20, 2012 at 7:54 AM, Frank Ch. Eigler <fche@redhat.com> wrote:
>>>> Hi -
>>>>
>>>> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>>>>> Something is wrong.  binutils git mirror hasn't been updated
>>>>> for several hours and there were 5 commits missing in
>>>>> git mirror.
>>>>
>>>> The way these mirrors are updated is by a rather CPU-expensive set of
>>>> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
>>>> fall behind.
>>>>
>>>
>>> 6 hours behind is very strange.
>>>
>>
>> Both gdb and binutils git mirrors are out of date.  Can someone
>> please take a look?
>
> There was one zombie mirror-sw process for each of those two
> repositories, and due to the new locking mechanism[*], no other job
> would proceed.  I've just killed all of the dependent processes
> and that let the zombies die for real, so now things seem to
> be moving again.
>

How big is a gap between CVS checkin and git mirror update?
> 1 hour?  There is a CVS checkin:

http://sourceware.org/ml/binutils-cvs/2012-11/msg00140.html

at 2012-11-21 13:53:23 UTC.  As of 2012-11-21 15:17:53 UTC,
git mirror hasn't been updated.


H.J.


-- 
H.J.

Re: Is git://sourceware.org/git/binutils.git down?

[Jim Meyering]

Frank Ch. Eigler wrote:

> Hi -
>
> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>> Something is wrong.  binutils git mirror hasn't been updated
>> for several hours and there were 5 commits missing in
>> git mirror.
>
> The way these mirrors are updated is by a rather CPU-expensive set of
> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
> fall behind.

Hi H.J.,

It looks like I've fixed it by doing this:

  rm -rf /home/meyering/mirror-git-to-cvs/.state/binutils/{.cvsps,find}

and letting the usual cron-driven command run again.
I see a commit in git from 4 hours ago.

    commit 84c2b3468b5ab55adf66c4fe787d5dbc91aa67c2
    Author: Alan Modra <amodra@bigpond.net.au>
    Date:   Wed Nov 21 23:00:04 2012 +0000

        daily update

Think of this as yet another good reason to move away from cvs.

Re: Is git://sourceware.org/git/binutils.git down?

[Christopher Faylor]

On Thu, Nov 22, 2012 at 04:22:24AM +0100, Jim Meyering wrote:
>Frank Ch. Eigler wrote:
>
>> Hi -
>>
>> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>>> Something is wrong.  binutils git mirror hasn't been updated
>>> for several hours and there were 5 commits missing in
>>> git mirror.
>>
>> The way these mirrors are updated is by a rather CPU-expensive set of
>> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
>> fall behind.
>
>Hi H.J.,
>
>It looks like I've fixed it by doing this:
>
>  rm -rf /home/meyering/mirror-git-to-cvs/.state/binutils/{.cvsps,find}
>
>and letting the usual cron-driven command run again.
>I see a commit in git from 4 hours ago.
>
>    commit 84c2b3468b5ab55adf66c4fe787d5dbc91aa67c2
>    Author: Alan Modra <amodra@bigpond.net.au>
>    Date:   Wed Nov 21 23:00:04 2012 +0000
>
>        daily update
>
>Think of this as yet another good reason to move away from cvs.

I guess.  But, it also sounds like a good reason to have some monitoring
tools available so that people who rely on git aren't left stranded for
days when the mirrors stop working.

cgf

Re: Is git://sourceware.org/git/binutils.git down?

[Jim Meyering]

Christopher Faylor wrote:
> On Thu, Nov 22, 2012 at 04:22:24AM +0100, Jim Meyering wrote:
>>Frank Ch. Eigler wrote:
>>
>>> Hi -
>>>
>>> On Tue, Nov 20, 2012 at 07:51:47AM -0800, H.J. Lu wrote:
>>>> Something is wrong.  binutils git mirror hasn't been updated
>>>> for several hours and there were 5 commits missing in
>>>> git mirror.
>>>
>>> The way these mirrors are updated is by a rather CPU-expensive set of
>>> scripts run by Jim Meyering.  I'm not too surprised if they sometimes
>>> fall behind.
>>
>>Hi H.J.,
>>
>>It looks like I've fixed it by doing this:
>>
>>  rm -rf /home/meyering/mirror-git-to-cvs/.state/binutils/{.cvsps,find}
>>
>>and letting the usual cron-driven command run again.
>>I see a commit in git from 4 hours ago.
>>
>>    commit 84c2b3468b5ab55adf66c4fe787d5dbc91aa67c2
>>    Author: Alan Modra <amodra@bigpond.net.au>
>>    Date:   Wed Nov 21 23:00:04 2012 +0000
>>
>>        daily update
>>
>>Think of this as yet another good reason to move away from cvs.
>
> I guess.  But, it also sounds like a good reason to have some monitoring
> tools available so that people who rely on git aren't left stranded for
> days when the mirrors stop working.

If someone wants to help, that'd be great.  I have a cron-driven script
already that was monitoring gdb cvs-vs-git diffs, but after more than
a year with no problems, I turned it off because the regular clone and
checkout did impose a cost on an already overburdened system.  It'd be
trivial to adapt to check binutils.

Maybe with more hardware headroom, or if someone wants to adjust things to
keep around a clone and cvs-checkout (where update cost would be minimal),
I can reenable it.