<div style="white-space:pre-wrap">Ceasar Sun, sorry that I forgot to add you in the list.<br><br>Here is the message.</div><br><div class="gmail_quote"><div dir="ltr">Adrian Reber <<a href="mailto:adrian@lisas.de">adrian@lisas.de</a>>於 2016年5月8日 週日，下午4:55寫道：<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">On Sun, May 08, 2016 at 02:58:05AM +0000, Cheng-Chia Tseng wrote:<br>
> I am one of the fedora ambassadors in Taiwan (an island next to mainland<br>
> China). NCHC, National Center for High-performance Computing, in Taiwan has<br>
> been mirroring fedora since 2015.<br>
><br>
> However, the local community found they could not get updates from NCHC for<br>
> some time and reported to me recently. It is not even listed in<br>
> <a href="https://admin.fedoraproject.org/mirrormanager/mirrors" rel="noreferrer" target="_blank">https://admin.fedoraproject.org/mirrormanager/mirrors</a> while it worked well<br>
> last year.<br>
><br>
> We are trying to figure out the possible cause and would like to fix it as<br>
> soon as possible. Any advice and suggestion is welcome. ;)<br>
><br>
> I also include the mirror manager in the discussion, we could work together<br>
> to get things done.<br>
<br>
I had a look at that host (<a href="http://free.nchc.org.tw" rel="noreferrer" target="_blank">free.nchc.org.tw</a>) and it was auto-disabled<br>
after 4 consecutive crawl failures. In this case the crawl took longer<br>
than three hours, which is the time we defined as maximum crawl time.<br>
<br>
I re-enabled the mirror and started a manual crawl to see if it the<br>
crawler now finishes within the three hour time limit.<br>
<br>
The easiest solution to avoid crawler timeouts is to provide a rsync URL<br>
which our crawler will use the scan the contents of the mirror. rsync is<br>
much better suited to scan the whole Fedora (and EPEL) directory than<br>
HTTP which is currently used to scan that mirror.<br>
<br>
Adrian<br>
--</blockquote></div>