Re: [Dovecot] Indexes.
Am 02.12.2010 14:53, schrieb Henrique Fernandes:
I have abou 9000 clients.
many clients
500 GB of storage used
small store
anyway you shouldnt run into problems with that
About the mail option at ocfs2 store creating, i guess the other guy that format the storage system did not use.
what a pitty , mail option help with small files performance like indexes and mail etc as far i remember
We are looking into ocfs2 1.6 that was out few months ago. But we are not able to get it running on or Centos5.5 yet.
hm centos is more redhat style i speculate Red Hat GFS would have been the better choice is there a list at centos/redhat/ocfs you might can asked about bugs with your ocfs2/distro combination ?
Are you talking abvout this optimizations?
Dovecot supports reading a few fields from the <base filename>: * ,S=<size>: <size> contains the file size. Getting the size from the filename avoids doing a stat(), which may improve the performance. This is especially useful with Maildir++ quota <http://wiki.dovecot.org/Quota/Maildir>. * ,W=<vsize>: <vsize> contains the file's RFC822.SIZE, ie. the file size with linefeeds being CR+LF characters. If the message was stored with CR+LF linefeeds, <size> and <vsize> are the same. Setting this may give a small speedup because now Dovecot doesn't need to calculate the size itself. A maildir filename with those fields would look something like: 1035478339.27041_118.foo.org <http://1035478339.27041_118.foo.org>,S=1000,W=1030:2,S
as far i remember keep close to nfs recommands
FS caching problems
NFS caching is a big problem when multiple computers are accessing the same mailbox simultaneously. The best fix for this is to prevent it from happening. Configure your setup so that a user always gets redirected to the same server (unless it's down). This also means that mail deliveries must be done by the same server, or alternatively it shouldn't update index files.
Dovecot flushes NFS caches when needed if you set mail_nfs_storage=yes, but unfortunately this doesn't work 100%, so you can get random errors.
Disabling NFS attribute cache helps a lot in getting rid of caching related errors, but this makes the performance MUCH worse and increases the load on NFS server. This can usually be done by giving actimeo=0 or noac mount option.
Index files
As described above, it's better to redirect users to the same server whenever possible. If you do this, it might also be a good idea to keep index files stored locally in that server. If user gets occasionally redirected to another server, the indexes will then be created locally there. This isn't a problem. However you might want to create a cronjob to delete old index directories.
If you choose to keep the index files stored in NFS, you'll need to set mmap_disable=yes. Both the mmap_disable and indexing to NFS will result in a notable performance hit. If you're not running lockd you'll have to set lock_method=dotlock, but this degrades performance. Note that many NFS installations have problems with lockd. If you're beginning to get all kinds of locking related errors, try if the problems go away with dotlocking.
but i agree, there should be more info with cluster file systems in the wiki
perhaps Timo has time to send recommands by mail, i am not sure that mine are optimal ( whatever i have no problems yet )
I will try send an image on the load of machine.
[]'sf.rique
On Thu, Dec 2, 2010 at 11:25 AM, Robert Schetterer
mailto:robert@schetterer.org> wrote: Am 02.12.2010 14:10, schrieb Henrique Fernandes: > With recomendation settings do you have ? http://wiki2.dovecot.org/MailLocation/SharedDisk and a few more optimizes about maildir > > how many clients ? i started the server not long ago so they are less yet ( about 100 maildirs ) but i have my own account in it which has massive mails ( 10 GB store with over 1000 mails daily in out , and perhaps 10000 mails in store) > > sorry, i did not say i use dovecot 2.0.6 looks fine, update to 2.0.8 if ready did you use the mail option at ocfs2 store creation ? what os/distro do you use , i remember bugs with ocfs with some distro versions > > []'sf.rique > > > On Thu, Dec 2, 2010 at 10:40 AM, Robert Schetterer > <robert@schetterer.org <mailto:robert@schetterer.org> <mailto:robert@schetterer.org <mailto:robert@schetterer.org>>> wrote: > > Am 02.12.2010 13:13, schrieb Henrique Fernandes: > > Hello people! > > > > I have huge problems with IO wait becase dovecot configured to use > maildir > > is under OCFS2 1.4. Now i have an question to OCFS2 each disk > action is > > really heavy becaue it has no index. > > > > Now i am thinking in what can be done to heltp my system to use > less disk. > > > > Looking for index and etc in dovecot i found this, this disables > the index > > file on disk and leave it on ram or it disable the index files ? > > > > If you really want to, you can also disable the index files > completely by > > appending :INDEX=MEMORY > > > > > > I am only trying to get dovecot to work faster. Right now we have > to waing > > some seconds to dovecot list all emails or read an email. > > > > I am also thinking in right index to an NFS but i don't know if > this makes > > any diferences in performance. > > > > Thanks to all! > > > > I use 3 servres wrinting on OCFS2 one is mailman, and other two > are to > > email, I have ipvs to load balance the conections. > > > > > > []'sf.rique > > > > hm , i have no problems with ocfs2 (1.4.3-1: amd64 ) on drbd ubuntu > lucid > using dovecot vers 2 recommended settings for cluster file systems > i have my index files in the maildir dir > > -- > Best Regards > > MfG Robert Schetterer > > Germany/Munich/Bavaria > > -- Best Regards MfG Robert Schetterer Germany/Munich/Bavaria
-- Best Regards
MfG Robert Schetterer
Germany/Munich/Bavaria
Did you see the IO wait in the picture i send ?
We did not use GFS cause we need an fencing hardware, they don't garante data without an fencing harware...
Anyway, we change to storage so we now are able to give a biger quota. Before use to be 200 mb and pretty much alluser uses pop.
We have 9000 users, but as i see not, only 4500 are actualy active.
Now, we are making teste with ocfs2 and ext3 we might gonna use NFS.
No, i don't have a list where i can ask! =/
Thanks for the help.
Can you see how many client you have log on at same time in you system ?
[]'sf.rique
On Thu, Dec 2, 2010 at 12:09 PM, Robert Schetterer robert@schetterer.orgwrote:
Am 02.12.2010 14:53, schrieb Henrique Fernandes:
I have abou 9000 clients.
many clients
500 GB of storage used
small store
anyway you shouldnt run into problems with that
About the mail option at ocfs2 store creating, i guess the other guy that format the storage system did not use.
what a pitty , mail option help with small files performance like indexes and mail etc as far i remember
We are looking into ocfs2 1.6 that was out few months ago. But we are not able to get it running on or Centos5.5 yet.
hm centos is more redhat style i speculate Red Hat GFS would have been the better choice is there a list at centos/redhat/ocfs you might can asked about bugs with your ocfs2/distro combination ?
Are you talking abvout this optimizations?
Dovecot supports reading a few fields from the <base filename>: * ,S=<size>: <size> contains the file size. Getting the size from the filename avoids doing a stat(), which may improve the performance. This is especially useful with Maildir++ quota <http://wiki.dovecot.org/Quota/Maildir>. * ,W=<vsize>: <vsize> contains the file's RFC822.SIZE, ie. the file size with linefeeds being CR+LF characters. If the message was stored with CR+LF linefeeds, <size> and <vsize> are the same. Setting this may give a small speedup because now Dovecot doesn't need to calculate the size itself. A maildir filename with those fields would look something like: 1035478339.27041_118.foo.org <http://1035478339.27041_118.foo.org>,S=1000,W=1030:2,S
as far i remember keep close to nfs recommands
FS caching problems
NFS caching is a big problem when multiple computers are accessing the same mailbox simultaneously. The best fix for this is to prevent it from happening. Configure your setup so that a user always gets redirected to the same server (unless it's down). This also means that mail deliveries must be done by the same server, or alternatively it shouldn't update index files.
Dovecot flushes NFS caches when needed if you set mail_nfs_storage=yes, but unfortunately this doesn't work 100%, so you can get random errors.
Disabling NFS attribute cache helps a lot in getting rid of caching related errors, but this makes the performance MUCH worse and increases the load on NFS server. This can usually be done by giving actimeo=0 or noac mount option.
Index files
As described above, it's better to redirect users to the same server whenever possible. If you do this, it might also be a good idea to keep index files stored locally in that server. If user gets occasionally redirected to another server, the indexes will then be created locally there. This isn't a problem. However you might want to create a cronjob to delete old index directories.
If you choose to keep the index files stored in NFS, you'll need to set mmap_disable=yes. Both the mmap_disable and indexing to NFS will result in a notable performance hit. If you're not running lockd you'll have to set lock_method=dotlock, but this degrades performance. Note that many NFS installations have problems with lockd. If you're beginning to get all kinds of locking related errors, try if the problems go away with dotlocking.
but i agree, there should be more info with cluster file systems in the wiki
perhaps Timo has time to send recommands by mail, i am not sure that mine are optimal ( whatever i have no problems yet )
I will try send an image on the load of machine.
[]'sf.rique
On Thu, Dec 2, 2010 at 11:25 AM, Robert Schetterer
mailto:robert@schetterer.org> wrote: Am 02.12.2010 14:10, schrieb Henrique Fernandes: > With recomendation settings do you have ? http://wiki2.dovecot.org/MailLocation/SharedDisk and a few more optimizes about maildir > > how many clients ? i started the server not long ago so they are less yet ( about 100 maildirs ) but i have my own account in it which has massive mails ( 10 GB store with over 1000 mails daily in out , and perhaps 10000 mails in store) > > sorry, i did not say i use dovecot 2.0.6 looks fine, update to 2.0.8 if ready did you use the mail option at ocfs2 store creation ? what os/distro do you use , i remember bugs with ocfs with some
distro
versions > > []'sf.rique > > > On Thu, Dec 2, 2010 at 10:40 AM, Robert Schetterer > <robert@schetterer.org <mailto:robert@schetterer.org> <mailto:robert@schetterer.org <mailto:robert@schetterer.org>>>
wrote:
> > Am 02.12.2010 13:13, schrieb Henrique Fernandes: > > Hello people! > > > > I have huge problems with IO wait becase dovecot configured to use > maildir > > is under OCFS2 1.4. Now i have an question to OCFS2 each
disk
> action is > > really heavy becaue it has no index. > > > > Now i am thinking in what can be done to heltp my system to
use
> less disk. > > > > Looking for index and etc in dovecot i found this, this disables > the index > > file on disk and leave it on ram or it disable the index
files ?
> > > > If you really want to, you can also disable the index files > completely by > > appending :INDEX=MEMORY > > > > > > I am only trying to get dovecot to work faster. Right now we have > to waing > > some seconds to dovecot list all emails or read an email. > > > > I am also thinking in right index to an NFS but i don't know
if
> this makes > > any diferences in performance. > > > > Thanks to all! > > > > I use 3 servres wrinting on OCFS2 one is mailman, and other
two
> are to > > email, I have ipvs to load balance the conections. > > > > > > []'sf.rique > > > > hm , i have no problems with ocfs2 (1.4.3-1: amd64 ) on drbd ubuntu > lucid > using dovecot vers 2 recommended settings for cluster file
systems
> i have my index files in the maildir dir > > -- > Best Regards > > MfG Robert Schetterer > > Germany/Munich/Bavaria > > -- Best Regards MfG Robert Schetterer Germany/Munich/Bavaria
-- Best Regards
MfG Robert Schetterer
Germany/Munich/Bavaria
On Thu, 2010-12-02 at 12:41 -0200, Henrique Fernandes wrote:
No, i don't have a list where i can ask! =/
https://www.redhat.com/mailman/listinfo/redhat-list is the more "general" list, its low traffic so you wont be inconvenienced by subscribing .
or maybe more specific... https://www.redhat.com/mailman/listinfo/linux-cluster (not on that list so I can't how helpful it would be.
participants (3)
-
Henrique Fernandes
-
Noel Butler
-
Robert Schetterer