- dovecot - dovecot.org

Re: [Dovecot] GPFS for mail-storage (Was: Re: Compressing existing maildirs)
by list＠airstreamcomm.net 04 Jan '12

04 Jan '12

Great information, thank you. Could you remark on GPFS services hosting mail storage over a WAN between two geographically separated data centers? ----- Reply message ----- From: "Jan-Frode Myklebust" <janfrode(a)tanso.net> To: "Stan Hoeppner" <stan(a)hardwarefreak.com> Cc: "Timo Sirainen" <tss(a)iki.fi>, <dovecot(a)dovecot.org> Subject: [Dovecot] GPFS for mail-storage (Was: Re: Compressing existing maildirs) Date: Tue, Jan 3, 2012 2:14 am On Sat, Dec 31, 2011 at 01:54:32AM -0600, Stan Hoeppner wrote: > Nice setup. I've mentioned GPFS for cluster use on this list before, > but I think you're the only operator to confirm using it. I'm sure > others would be interested in hearing of your first hand experience: > pros, cons, performance, etc. And a ball park figure on the licensing > costs, whether one can only use GPFS on IBM storage or if storage from > others vendors is allowed in the GPFS pool. I used to work for IBM, so I've been a bit uneasy about pushing GPFS too hard publicly, for risk of being accused of being biased. But I changed job in November, so now I'm only a satisfied customer :-) Pros: Extremely simple to configure and manage. Assuming root on all nodes can ssh freely, and port 1191/tcp is open between the nodes, these are the commands to create the cluster, create a NSD (network shared disks), and create a filesystem: # echo hostname1:manager-quorum > NodeFile # "manager" means this node can be selected as filesystem manager # echo hostname2:manager-quorum >> NodeFile # "quorum" means this node has a vote in the quorum selection # echo hostname3:manager-quorum >> NodeFile # all my nodes are usually the same, so they all have same roles. # mmcrcluster -n NodeFile -p $(hostname) -A ### sdb1 is either a local disk on hostname1 (in which case the other nodes will access it over tcp to ### hostname1), or a SAN-disk that they can access directly over FC/iSCSI. # echo sdb1:hostname1::dataAndMetadata:: > DescFile # This disk can be used for both data and metadata # mmcrnsd -F DescFile # mmstartup -A # starts GPFS services on all nodes # mmcrfs /gpfs1 gpfs1 -F DescFile # mount /gpfs1 You can add and remove disks from the filesystem, and change most settings without downtime. You can scale out your workload by adding more nodes (SAN attached or not), and scale out your disk performance by adding more disks on the fly. (IBM uses GPFS to create scale-out NAS solutions http://www-03.ibm.com/systems/storage/network/sonas/ , which highlights a few of the features available with GPFS) There's no problem running GPFS on other vendors disk systems. I've used Nexsan SATAboy earlier, for a HPC cluster. One can easily move from one disksystem to another without downtime. Cons: It has it's own page cache, staticly configured. So you don't get the "all available memory used for page caching" behaviour as you normally do on linux. There is a kernel module that needs to be rebuilt on every upgrade. It's a simple process, but it needs to be done and means we can't just run "yum update ; reboot" to upgrade. % export SHARKCLONEROOT=/usr/lpp/mmfs/src % cp /usr/lpp/mmfs/src/config/site.mcr.proto /usr/lpp/mmfs/src/config/site.mcr % vi /usr/lpp/mmfs/src/config/site.mcr # correct GPFS_ARCH, LINUX_DISTRIBUTION and LINUX_KERNEL_VERSION % cd /usr/lpp/mmfs/src/ ; make clean ; make World % su - root # export SHARKCLONEROOT=/usr/lpp/mmfs/src # cd /usr/lpp/mmfs/src/ ; make InstallImages > > To this point IIRC everyone here doing clusters is using NFS, GFS, or > OCFS. Each has its downsides, mostly because everyone is using maildir. > NFS has locking issues with shared dovecot index files. GFS and OCFS > have filesystem metadata performance issues. How does GPFS perform with > your maildir workload? Maildir is likely a worst case type workload for filesystems. Millions of tiny-tiny files, making all IO random, and getting minimal controller read cache utilized (unless you can cache all active files). So I've concluded that our performance issues are mostly design errors (and the fact that there were no better mail storage formats than maildir at the time these servers were implemented). I expect moving to mdbox will fix all our performance issues. I *think* GPFS is as good as it gets for maildir storage on clusterfs, but have no number to back that up ... Would be very interesting if we could somehow compare numbers for a few clusterfs'. I believe our main limitation in this setup is the iops we can get from the backend storage system. It's hard to balance the IO over enough RAID arrays (the fs is spread over 11 RAID5 arrays of 5 disks each), and we're always having hotspots. Right now two arrays are doing <100 iops, while others are doing 4-500 iops. Would very much like to replace it by something smarter where we can utilize SSDs for active data and something slower for stale data. GPFS can manage this by itself trough it's ILM interface, but we don't have the very fast storage to put in as tier-1. -jf

2 1

[Dovecot] Problem with huge IMAP Archive after Courier migration
by Preacher 04 Jan '12

04 Jan '12

I have a mail server running Debian 6.0 with Courier IMAP to store project related mail. Currently the maildir of the archive (one user) contains about 37GB of data. Our staff is acessing the archive via Outlook 2007 where they drag their Exchange inbox or sent files to it. The problem with courier is that is sometimes mixes up headers with message bodies, so I wanted to migrate to dovecot. I tried this on my proxy running Debian 7.0 with some test data and this worked fine (OK, spent some hours to get the config files done - Dovecot without authentication). Dovecot version here is 2.0.15. Tried it with our productive system today, but got Dovecot 1.2.15 installed on Debian 6.0 Config files and parameters I took from my test system were not compatible and I didn't get it to work. So I forced to install the Debisn 7.0 packages with 2.0.15 and finally got the server running, I also restarted the whole machine to empty caches. But the problem I got was that in the huge folder hierarchy the downloaded headers in the individual folders disappeared, some folders showed a few very old messages, some none. Also some subfolders disappeared. I checked this with Outlook and Thunderbird. The difference was, that Thunderbird shows more messages (but not all) than Outlook in some folders, but also none in some others. Outlook brought up a message in some cases, that the connection timed out, although I set the timeout to 60s. After being frustrated uninstalled dovecot, went back to Courier and folder contents are displayed correctly again. Anyone a clue what's wrong here? Finally some config information: proxy-server:~# dovecot -n # 2.0.15: /etc/dovecot/dovecot.conf # OS: Linux 3.1.0-1-686-pae i686 Debian wheezy/sid auth_debug_passwords = yes auth_mechanisms = plain login disable_plaintext_auth = no namespace { inbox = yes location = prefix = INBOX. separator = . type = private } passdb { driver = pam } plugin { sieve = ~/.dovecot.sieve sieve_dir = ~/sieve } protocols = imap ssl = no ssl_cert = </etc/ssl/certs/dovecot.pem ssl_key = </etc/ssl/private/dovecot.pem userdb { driver = passwd }

5 7

[Dovecot] Maildir migration and uids
by David Jonas 03 Jan '12

03 Jan '12

I'm in the process of migrating a large number of maildirs to a 3rd party dovecot server (from a dovecot server). Tests have shown that using imap to sync the accounts doesn't preserve the uidl for pop3 access. My current attempt is to convert the maildir to mbox and add an X-UIDL header in the process. Run a second dovecot that serves the converted mbox. But dovecot's docs say, "None of these headers are sent to IMAP/POP3 clients when they read the mail". Is there any way to sync these maildirs to the new server and maintain the uids? The real goal is keep customers email clients happy when they are pointed at the new server. Am I just wishing?

2 2

[Dovecot] Deliver all addresses to the same mdbox:?
by Ralf Hildebrandt 03 Jan '12

03 Jan '12

For archiving purposes I'm delivering all addresses to the same mdbox: like this: passdb { driver = passwd-file args = username_format=%u /etc/dovecot/passwd } userdb { driver = static args = uid=1000 gid=1000 home=/home/copymail allow_all_users=yes } Yet I'm getting this: Jan 3 19:03:27 mail postfix/lmtp[29378]: 3THjg02wfWzFvmL: to=<firstname.lastname@charite.de@backup.invalid>, relay=mail.charite.de[private/dovecot-lmtp], conn_use=20, delay=323, delays=323/0/0/0, dsn=4.1.1, status=SOFTBOUNCE (host mail.charite.de[private/dovecot-lmtp] said: 550 5.1.1 <"firstname.lastname(a)charite.de"@backup.invalid> User doesn't exist: "firstname.lastname(a)charite.de"@backup.invalid (in reply to RCPT TO command)) (using soft_bounce = yes here in Postfix) In short: backup.invalid is delivered to dovecot by means of LMTP (local socket). I thought my static mapping in userdb would enable the lmtp listener to accept ALL recipients and map their $home to /home/copymail - why is that not working? -- Ralf Hildebrandt Geschäftsbereich IT | Abteilung Netzwerk Charité - Universitätsmedizin Berlin Campus Benjamin Franklin Hindenburgdamm 30 | D-12203 Berlin Tel. +49 30 450 570 155 | Fax: +49 30 450 570 962 ralf.hildebrandt(a)charite.de | http://www.charite.de

2 2

[Dovecot] Multiple Maildirs per Virtual User
by Ruslan Nabioullin 03 Jan '12

03 Jan '12

How would it be possible to configure dovecot (2.0.16) in such a way that it would serve several maildirs (e.g., INBOX, INBOX.Drafts, INBOX.Sent, forum_email, [Gmail].Trash, etc.) per virtual user? I am only able to specify a single maildir, but I want all maildirs in /home/my-username/mail/account1/ to be served. e.g., /etc/dovecot/passwd: my-username_account1:{PLAIN}password:my-username:my-group::::userdb_mail=maildir:/home/my-username/mail/account1/INBOX Thanks in advance, Ruslan -- Ruslan Nabioullin rnabioullin(a)gmail.com

2 3

[Dovecot] What is normal CPU usage of dovecot imap?
by Mikko Lampikoski 03 Jan '12

03 Jan '12

I got Dual Core Intel Xeon CPU 3.00GHz, over 1000 mailbox and almost 1 dovecot login / second (peak time). Server stats says that load is continually over 2 and cpu usage is 60%. top says that imap is making this load. virtual users are in mysql database and mysqld is running on another server (this server is ok). Do I need better CPU or is there something going on that I do not understand? # dovecot -n # 1.1.11: /etc/dovecot/dovecot.conf # OS: Linux 2.6.32-4-pve i686 Ubuntu 9.10 nfs log_timestamp: %Y-%m-%d %H:%M:%S protocols: imap imaps pop3 pop3s listen: *, [::] ssl_ca_file: /etc/ssl/**********.crt ssl_cert_file: /etc/ssl/**********.crt ssl_key_file: /etc/ssl/**********.key ssl_key_password: ********** disable_plaintext_auth: no verbose_ssl: yes shutdown_clients: no login_dir: /var/run/dovecot/login login_executable(default): /usr/lib/dovecot/imap-login login_executable(imap): /usr/lib/dovecot/imap-login login_executable(pop3): /usr/lib/dovecot/pop3-login login_greeting_capability(default): yes login_greeting_capability(imap): yes login_greeting_capability(pop3): no login_process_size: 128 login_processes_count: 10 login_max_processes_count: 2048 mail_max_userip_connections(default): 10 mail_max_userip_connections(imap): 10 mail_max_userip_connections(pop3): 3 first_valid_uid: 99 last_valid_uid: 100 mail_privileged_group: mail mail_location: maildir:/var/vmail/%d/%n:INDEX=/var/indexes/%d/%n fsync_disable: yes mail_nfs_storage: yes mbox_write_locks: fcntl mbox_min_index_size: 4 mail_executable(default): /usr/lib/dovecot/imap mail_executable(imap): /usr/lib/dovecot/imap mail_executable(pop3): /usr/lib/dovecot/pop3 mail_process_size: 2048 mail_plugin_dir(default): /usr/lib/dovecot/modules/imap mail_plugin_dir(imap): /usr/lib/dovecot/modules/imap mail_plugin_dir(pop3): /usr/lib/dovecot/modules/pop3 imap_client_workarounds(default): outlook-idle imap_client_workarounds(imap): outlook-idle imap_client_workarounds(pop3): pop3_client_workarounds(default): pop3_client_workarounds(imap): pop3_client_workarounds(pop3): outlook-no-nuls auth default: mechanisms: plain login cram-md5 cache_size: 1024 passdb: driver: sql args: /etc/dovecot/dovecot-sql.conf userdb: driver: static args: uid=99 gid=99 home=/var/vmail/%d/%n allow_all_users=yes socket: type: listen client: path: /var/spool/postfix/private/auth-client mode: 432 user: postfix group: postfix master: path: /var/run/dovecot/auth-master mode: 384 user: vmail

2 3

[Dovecot] Compressing existing maildirs
by Jan-Frode Myklebust 03 Jan '12

03 Jan '12

I've just enabled zlib for our users, and am looking at how to compress the existing files. The routine for doing this at http://wiki2.dovecot.org/Plugins/Zlib seems a bit complicated. What do you think about simply doing: find /var/vmail -type f -name "*,S=*" -mtime +1 -exec gzip -S Z -6 '{}' + I.e. find all maildir-files: - with size in the name ("*,S=*") - modified before I enabled zlib plugin - compress them - add the Z suffix - keep timestamps (gzip does that by default) It's of course racy without the maildirlock, but are there any other problems with this approach ? -jf

4 14

[Dovecot] Dsync fails on second sync for folders with dot in the name
by Jan-Frode Myklebust 03 Jan '12

03 Jan '12

I'm in the processes of running our first dsync backup of all users (from maildir to mdbox on remote server), and one problem I'm hitting that dsync will work fine on first run for some users, and then reliably fail whenever I try a new run: $ sudo dsync -u janfrode(a)example.net backup ssh -q mailbackup(a)repo1.example.net dsync -u janfrode(a)example.net $ sudo dsync -u janfrode(a)example.net backup ssh -q mailbackup(a)repo1.example.net dsync -u janfrode(a)example.net dsync-remote(janfrode(a)example.net): Error: Can't delete mailbox directory INBOX/a: Mailbox has children, delete them first The problem here seems to be that this user has a maildir named ".a.b". On the backup side I see this as "a/b/". So dsync doesn't quite seem to agree with itself for how to handle folders with dot in the name. -jf

3 7

[Dovecot] lmtp-postlogin ?
by Jan-Frode Myklebust 03 Jan '12

03 Jan '12

We have last-login tracking for imap and pop, and I intend to use this for deciding which users to backup daily. But, it would also be nice to backup users who has only received messages, but not logged in lately.. So is it possible to implement "last-login" tracking for lmtp ? I naively tried copying the settings from imap, but it didn't work: service lmtp-postlogin { executable = script-login /usr/local/sbin/lmtp-postlogin.sh unix_listener lmtp-postlogin { } } service lmtp { executable = lmtp lmtp-postlogin <snip> -jf

2 3

[Dovecot] Newbie: LDA Isn't Logging
by Michael Papet 03 Jan '12

03 Jan '12

Hi, I'm a newbie having some trouble getting deliver to log anything. Related to this, there are no return values unless the -d is missing. I'm using LDAP to store virtual domain and user account information. Test #1: /usr/lib/dovecot/deliver -e -f mpapet(a)yahoo.com -d zed(a)mailswansong.dom < bad.mail Expected result: supposed to fail, there's no zed account via ldap lookup and supposed to get a return code per the wiki at http://wiki2.dovecot.org/LDA. Supposed to log too. Actual result: nothing gets delivered, no return code, nothing is logged. Test #2: /usr/lib/dovecot/deliver -f mpapet(a)yahoo.com -d dude(a)mailswansong.dom < good.mail Expected result: deliver to dude and return 0. Actual result: delivers, but no return code. Nothing logged. The wiki is vague about the difficulties of getting deliver LDA to log, but I thought I had it covered in my config. I even opened permissions up wide (777) on my log files specified below. Nothing gets logged. The ONLY thing changed in 15-lda.conf is as follows. protocol lda { # Space separated list of plugins to load (default is global mail_plugins). #mail_plugins = $mail_plugins log_path = /var/log/dovecot/lda.log info_log_path = /var/log/dovecot/lda-info.log service auth { unix_listener auth-client { mode = 0600 user = vmail } } I'm running plain Debian Testing and used dovecot from Debian's repository. The end-goal is to write a qpsmtpd queue plugin, but I need to figure out what's the matter first. Thanks in advance. mpapet

2 1