[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Alioth tracker



On Sun, 2014-05-11 at 21:38 +0200, Tollef Fog Heen wrote:
> I'm not disagreeing, I think we're providing a much poorer service level
> for Alioth than what we should do.  Sadly, I don't have the motivation
> to spend much time there nowadays.

I have for years hosted my own projects in a minimalistic fashion [0],
and as a consequence have been nagged to provide "modern amenities" like
issue trackers and DCVS.  The obvious solution would be to move to free
hosting like sourceforge, but sourceforge is closed source.  Then I
joined Debian.  Alioth seemed like a natural fit, so I started moving my
projects to it.

But now I've decided that's not such a good idea.  Not because of some
of the issues mentioned in this thread - like DVCS support of lacking
features in Fusion Forge.  I can work around them.

My problem is Alioth isn't reliable enough.  In week or so I used it:

a.  As Ole mentioned, there are 180 odd open support requests, dating
    back 7 years.  It's not that things aren't being done - Stephen Gran
    in particular appears to regularly attend to the list and close
    issues.  However, there should be no support requests open for more
    than a few weeks (and ideally at most a couple of days).  Many of
    these old "support requests" are bugs or features, and there are
    separate trackers for those.  In other words, the support list needs
    to be triaged.  If after triage there are still support requests
    more than a month old, the clearly Alioth needs extra admin
    manpower.  Right now it is difficult to tell if manpower is really
    the issue.

b.  For a while when I was using it is was horribly slow (as in taking
    minutes to send a response to a HTTP request).  I could not see
    why.  After a day or so the issue went away and it because usable
    again.

c.  Then I started getting mysterious failures.  After a bit of digging
    around I noticed /tmp was at 100%.  Someone fixed this after a few
    hours.

d.  At the same time I noticed disk space it is sitting 94% usage.
    The amount of disk used is under 600GB.

e.  I suspect running out of disk space on /tmp caused a number of
    other issues for me [1].  The details aren't relevant here.  What is
    relevant is in order to diagnose what was going I poked around the
    file system, and noticed a number of other much older projects were
    suffering from the same issues.  Since this means among other things
    they can't use the DVCS, presumably they had been abandoned.

f.  After seeing all this, I decided I had better do some "due
    diligence" and what backup arrangements were in place.  As far as
    I can tell there aren't any.

At this point I reluctantly decided I had to use why I was trying to
avoid - a commercial provider running close source.

If three things changed on Alioth I would move back.  They are:

A.  Solve the disk space problems.
B.  A backup system.
C.  Support list triaged, and it's length viewed as a KPI.

If the Alioth team thinks I could be useful in getting this things done,
I'd be happy to become part of it.  Even if they don't, I'd be happy to
donate 2 x 4TB drives so the disk space issue can be fixed - assuming
there are remote hands available to fit them.

[0]  http://www.stuart.id.au/russell/files
[1]
https://alioth.debian.org/tracker/index.php?func=detail&aid=314680&group_id=1&atid=200001

Attachment: signature.asc
Description: This is a digitally signed message part


Reply to: