archiva-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Julien CARSIQUE <>
Subject Re: Archiva freeze because of down site ?
Date Fri, 21 Dec 2007 13:46:52 GMT
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
  <meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<body bgcolor="#ffffff" text="#000000">
Thanks, good news.<br>
So I'm waiting impatiently for the 1.0.1 :-)<br>
Brett Porter a &eacute;crit&nbsp;:
  <pre wrap="">Yeah, improvements to the caching should certainly be filed as a
feature request - and we should run through the plan on the dev@ list
if possible.

As far s your original freezing problem goes - I assigned it to the
1.0.1 version yesterday. I've seen another report too and they are
helping to diagnose what the cause is. I had a suspicion it might be
because of remote repository connections - I think this evidence is
starting to confirm that.

- Brett

On 21/12/2007, nicolas de loof <a class="moz-txt-link-rfc2396E" href="">&lt;;</a>
  <blockquote type="cite">
    <pre wrap="">"Cache failure" option is supposed (AFAIK) to enable such caching,
and avoid
requesting again and again the same repository, where 404 or timeout occurs.

Reading the code, (DefaultRepositoryProxyConnectors) seems the cache is not
used before requesting a proxyConnector for an artifact. 404 also are not
cached, and IMHO they should for performance consideration (exemple : many
artifact have no -sources.jar, but are requested by IDE plugins).

The cache configuration is also hardcoded in archiva-policies. Not sure it
will fit the requirements for all users and use-cases (missing artifact,
network connection borken ...).

I'll look at this tomorow, could you please fill an issue in Jira ?


2007/12/20, Julien CARSIQUE <a class="moz-txt-link-rfc2396E" href="">&lt;;</a>:
    <blockquote type="cite">
      <pre wrap="">Hello,

Today, archiva was (again freezed) but restart didn't solve the problem.
Looking in the logs, I saw that one of our remote repositories was dead;
removing it solved the problem.
Archiva do not cache information about unavailable site to not calling
it for each artifact ?
Could be a useful improvement ?

PS: I didn't have any answer about my mail "Cannot use archiva browse
function after 2 days running"; we still have to restart archiva every
one, two or three days because it freezes. Please, help would be very
welcome; thanks.

This morning, Archiva was freezed again; here's the top on the server :
25619 archiva   19   0      209m 186m 103m R 17.0     21.1      53:14.14

25751 archiva   20   0      209m 186m 103m R 16.8     21.1      47:34.28

28163 archiva   14   0      209m 186m 103m S 16.6     21.1      40:42.57

28124 archiva   14   0      209m 186m 103m S 16.4     21.1      41:44.36

26348 archiva   19   0      209m 186m 103m R 15.7     21.1      43:01.91

Julien CARSIQUE, Nuxeo (Paris, France)
Open Source Enterprise Content Management - <a class="moz-txt-link-freetext" href=""></a>
Nuxeo EP 5: extensible, Java EE and standards based ECM Platform
<a class="moz-txt-link-freetext" href=""></a>
- Tel: +33 1 40 33 79 87

  <pre wrap=""><!---->


View raw message