Commit Graph

117 Commits

Author SHA1 Message Date
Adrian Reber
d7a367ae66 Decrease the number of parallel crawlers from 35 to 32 2015-06-12 16:11:00 +00:00
Adrian Reber
7cfe45d7d8 Revert "Added hotfix to not disable always up2date categories."
This reverts commit 4fd68994e1.
2015-06-08 08:43:48 +00:00
Kevin Fenzi
f938d7df5e Try lowering it some more, as mm-frontend01 keeps hitting OOM. 2015-06-01 16:32:08 +00:00
Kevin Fenzi
696ac2ea09 Try swapping this around and doing more processes with less threads. 2015-06-01 14:41:58 +00:00
Kevin Fenzi
423d76183d Try moving this down to 5 2015-06-01 13:08:53 +00:00
Kevin Fenzi
ed69abc4c0 Try upping threads here on mirrormanager frontend to see if it can cope with request spikes better. 2015-06-01 12:17:58 +00:00
Kevin Fenzi
d1621a6f3b Need to actually listen on 443 somewhere, this is as good a place as any 2015-05-31 16:44:12 +00:00
Adrian Reber
af9462caab Reduce number of parallel crawlers from 38 to 35
To avoid regular nagios memory warnings the number of parallel crawlers
is reduced from 38 to 35 to reduce the required memory.
2015-05-31 08:11:04 +00:00
Patrick Uiterwijk
81e0f3986f Add nagios internal monitoring for mm2 publiclist and fix mm2 static alias
Signed-off-by: Patrick Uiterwijk <puiterwijk@redhat.com>
2015-05-20 15:20:58 +00:00
Adrian Reber
7c2aefb8ad Add rewrite rule to access specific parts of the mirrorlist
The MM1 publiclist links are supposed to still work. If an old
publiclist link tries to directly access a product/version or
product/version/arch combination an additional 'mirrors' needs to be
inserted into the link. For full MM1 compatibility following PR is
necessary:

https://github.com/fedora-infra/mirrormanager2/pull/76
2015-05-19 07:45:20 +00:00
Pierre-Yves Chibon
01c667be07 Tag the mirrormanager2 frontend tasks 2015-05-18 11:32:02 +02:00
Pierre-Yves Chibon
45d8e7e0d7 Configure fedmenu for mirrormanager2 2015-05-18 11:30:36 +02:00
Adrian Reber
74c772f99f Fix redirect from publiclist to mirrormanager.
There have been reports that the old links pointing to the mirrorlist
are no longer working:

 $ curl -I http://mirrors.fedoraproject.org/publiclist/
 Location: https://admin.fedoraproject.org/mirrormanager///

This redirect still works but trying to access a specific mirrorlist
fails:

 $ curl -I http://mirrors.fedoraproject.org/publiclist/EPEL/7/
 Location: https://admin.fedoraproject.org/mirrormanager///EPEL/7/
 $ curl -I https://admin.fedoraproject.org/mirrormanager///EPEL/7/
 HTTP/1.1 404 NOT FOUND

At different places there are just too many slashes added. Removing the
slashes seems to help.
2015-05-15 19:06:05 +00:00
Adrian Reber
48e181ab24 Also crawl the last mirror in the database.
The last mirror in the database was not crawled and this adds '1' to
the --stopid if necessary.
2015-05-14 08:48:59 +00:00
Adrian Reber
2fc7977b35 Increase crawler timeout from 2h to 3h.
Since MM2 is in production about 140 mirrors have been auto-disabled
due to crawler timing out after 2 hours (default). Try if it works
better with 3 hours. This in combination with the previous commit
to decrease the load on the database should help to auto disable
less good mirrors. Especially mirrors who mirroring almost
everything can hardly be crawled within the 2 hour limit. Unfortunately
the limit is per host and not category.
2015-05-14 08:48:58 +00:00
Adrian Reber
703a46bada Start the crawl later on the second crawler.
Even with rsync as crawl method some hosts are taking a very long time
to be crawled. The network connection with rsync is only open for a
short time, but with both crawlers reading and writing from the database
it takes a very long time until the status of all directories is
updated. Therefore this patch introduces a 3 hour delay of the crawl
on the second crawler. This could also be solved with two different
cron.d files; one for each crawler.
2015-05-14 08:48:58 +00:00
Adrian Reber
66ae58a98b Document pull request for this hotfix 2015-05-08 20:25:23 +00:00
Adrian Reber
4fd68994e1 Added hotfix to not disable always up2date categories. 2015-05-08 20:14:00 +00:00
Adrian Reber
41e821f9f7 Do not update backend cron script in staging 2015-05-07 12:34:54 +00:00
Adrian Reber
02f5f3e86a Sync MM2 pickles to the mirrorlist servers after they have been re-created. 2015-05-07 10:43:33 +00:00
Adrian Reber
fa9ac723f6 Start httpd on role frontend2 and not frontend 2015-05-07 10:32:55 +00:00
Adrian Reber
57117196cc Enable and start httpd on mm-crawler and mm-frontend 2015-05-07 10:19:37 +00:00
Adrian Reber
0d28a56601 The mirrorlist server expects i2_netblocks.txt and not internet2_netblocks.txt 2015-05-07 10:19:37 +00:00
Adrian Reber
35296c746d Only sync the files actually needed on the mirrorlist servers
Also enable rsync compression like the old script did.
2015-05-07 10:19:37 +00:00
Adrian Reber
7796169f6b Setup /var/lib/mirrormanager/.ssh directory for mm-backend 2015-05-07 08:03:15 +00:00
Adrian Reber
7ead116450 Enable redirect for the new MM2 publiclist 2015-05-07 07:37:42 +00:00
Adrian Reber
d8e91732e7 Add script to sync MM2 generated pkls to the mirrorlist servers 2015-05-07 07:20:58 +00:00
Adrian Reber
de58ca2f0f Forgot to change the actual command to run 38 crawlers in parallel 2015-05-06 17:28:32 +00:00
Adrian Reber
cb8716913e Remember the start time of the last umdl run
The script which checks if umdl should be run used to store
the time of the end of the current umdl run. It actually should
remember the time of the start of this run as we want to make sure
to pick up all changes since the start of the current run.

Also print the date and time in the same line as the logging.
2015-05-06 12:58:12 +00:00
Adrian Reber
c7f31e41bb Increase the number of parallel crawls from 35 to 38. 2015-05-06 12:36:06 +00:00
Adrian Reber
3357ba0d37 Do not automatically run the crawler in staging 2015-05-06 12:34:08 +00:00
Adrian Reber
16e0082824 Distribute the hosts to be crawled by MM2 between the existing crawlers
This commit adds two new scripts to the crawlers:
 * mm2_get-highest-active-host-id: this queries the MM database and returns
   the id of the active mirror with the highest id
 * run_crawler.sh: this calculates the right startid and stopid parameters
   depending on the number of available crawlers

These scripts are integrated into the start of the crawler so that each
crawler only crawls a subset of the existing mirrors. This distribution
is not perfect as it pretends that the active mirrors are equally
distributed under the total number of existing mirrors.
2015-05-06 09:08:08 +00:00
Kevin Fenzi
e6f8efe736 Change this alias to mirrormanager as thats the url people know and use already. 2015-05-05 16:36:52 +00:00
Pierre-Yves Chibon
c491c0a141 Fix indentation 2015-05-05 13:09:36 +02:00
Pierre-Yves Chibon
57d6053166 Apparently quotes work 2015-05-05 13:08:02 +02:00
Pierre-Yves Chibon
5d0c06fc22 Try with quotes 2015-05-05 12:52:04 +02:00
Adrian Reber
0b6c4ac699 move status message to the right location 2015-05-05 08:44:44 +00:00
Pierre-Yves Chibon
07cd4b2c3c Add and install the alembic.ini for MM2 2015-05-04 22:32:58 +02:00
Adrian Reber
69348cdbab installed wrong script; replaced with the correct script 2015-05-04 12:20:59 +00:00
Adrian Reber
3e140185bd Only run umdl if the data has actually changed
/usr/bin/mm2_update-master-directory-list is now wrapped with a
script which queries the fedmsg bus if there is actually new data
on the master mirror. So now it is wrapped with a lock wrapper
and fedmsg query wrapper.
2015-05-04 10:20:45 +00:00
Kevin Fenzi
e0b72c224b Put things back for mm1 publiclists until mm2 is live. 2015-04-24 23:01:59 +00:00
Adrian Reber
a16f9992be Run the script for Internet2 netblocks very late
to make sure the files are all there. The script is
looking in date dependent directories and depending
on the timezone of the remote server the files appear late.
2015-04-24 08:51:09 +00:00
Adrian Reber
a845b5df35 Make the MM2 backend scripts also run as mirrormanager (instead as root). 2015-04-24 07:53:55 +00:00
Kevin Fenzi
6ed257f73f Redirect over to mm frontend for publiclist and / on mirrors 2015-04-23 20:26:33 +00:00
Adrian Reber
0b608dd24a Run the crawler as mirrormanager and not as root. 2015-04-23 19:12:15 +00:00
Adrian Reber
c68f132427 Update MM2 configuration file to match installed MM2 components 2015-04-23 10:36:37 +00:00
Adrian Reber
de9eff0a44 32GB of RAM is not enough for 75 threads, 40 seems to work so far 2015-04-23 07:56:31 +00:00
Adrian Reber
0dad6bee86 Apply selinux type to crawler log files 2015-04-23 07:48:20 +00:00
Adrian Reber
e0ba0f7c74 32GB of RAM is not enough for 75 threads; let's try with 50 threads 2015-04-15 12:51:10 +00:00
Patrick Uiterwijk
fb7e6e541b Add a Location tag for static
I got the second +1, so let's re-enable the patch

This reverts commit 0bdeabc8d4.
2015-04-14 18:57:20 +00:00