Lag in new hackage mirroring of old hackage
Hello, I want to report some lag in the mirroring of old hackage by the new hackage. "Note: we are mirroring packages from the old server (currently every 30min) so it is suitable to use as your main hackage server with some caveats: because we are allowing package authors to upload (as well as the mirroring) then you may find a slightly different set of packages on this server. " -- http://new-hackage.haskell.org/ viewed on August 2, 2013 But for some elm packages, I just noticed today, the versions on new hackage are considerably more than 30 minutes older than those on old hackage: Elm: new hackage 0.7.1.1 uploaded January 25, 2013 old hackage 0.8.0.3 uploaded May 28, 2013 elm-server: new hackage 0.7 uploaded January 4, 2013 old hackage 0.8 uploaded May 22, 2013 I've been happily using the new hackage for some weeks now, and occasionally enjoying the new features. Other than this lag, it's been a good experience. Thanks! -- Gregory D. Weber, Ph. D. http://mypage.iu.edu/~gdweber/ Associate Professor of Informatics Tel (765) 973-8420 Indiana University East FAX (765) 973-8550
On Thu, 2013-08-08 at 16:31 -0400, Gregory Weber wrote:
Hello,
I want to report some lag in the mirroring of old hackage by the new hackage.
"Note: we are mirroring packages from the old server (currently every 30min) so it is suitable to use as your main hackage server with some caveats: because we are allowing package authors to upload (as well as the mirroring) then you may find a slightly different set of packages on this server. "
-- http://new-hackage.haskell.org/ viewed on August 2, 2013
But for some elm packages, I just noticed today, the versions on new hackage are considerably more than 30 minutes older than those on old hackage:
I realise this is a very late reply, but a little explanation is perhaps still interesting. Yes, the mirroring did fail and we didn't have any monitoring set up or auto restart so we didn't notice for a while. I think I have identified and fixed the cause of the failure (the mirror client was designed to be robust to mirroring problems but this one slipped through). On the new beta site the mirror client is running under an init script and the infrastructure team are going to sort out auto restart and monitoring for us. -- Duncan Coutts, Haskell Consultant Well-Typed LLP, http://www.well-typed.com/
Thanks for your attention to this. Late reply is better than none, and it was not a burning issue, since it was easy to switch back to the non-new hackage. Greg On 2013-Sep-11, Duncan Coutts and/or a Mail User Agent wrote:
On Thu, 2013-08-08 at 16:31 -0400, Gregory Weber wrote:
Hello,
I want to report some lag in the mirroring of old hackage by the new hackage.
"Note: we are mirroring packages from the old server (currently every 30min) so it is suitable to use as your main hackage server with some caveats: because we are allowing package authors to upload (as well as the mirroring) then you may find a slightly different set of packages on this server. "
-- http://new-hackage.haskell.org/ viewed on August 2, 2013
But for some elm packages, I just noticed today, the versions on new hackage are considerably more than 30 minutes older than those on old hackage:
I realise this is a very late reply, but a little explanation is perhaps still interesting.
Yes, the mirroring did fail and we didn't have any monitoring set up or auto restart so we didn't notice for a while. I think I have identified and fixed the cause of the failure (the mirror client was designed to be robust to mirroring problems but this one slipped through).
On the new beta site the mirror client is running under an init script and the infrastructure team are going to sort out auto restart and monitoring for us.
-- Duncan Coutts, Haskell Consultant Well-Typed LLP, http://www.well-typed.com/
-- Gregory D. Weber, Ph. D. http://mypage.iu.edu/~gdweber/ Associate Professor of Informatics Tel (765) 973-8420 Indiana University East FAX (765) 973-8550
participants (3)
-
Duncan Coutts -
Gregory D. Weber -
Gregory Weber