Re: replicator: Panic: data stack: Out of memory when allocating 268435496 bytes

Hello,

try updating to Dovecot v2.3.20 // I had the same in earlier versions, but it doesn't happen any longer now.

Increasing the limit didn't help me at that time, even the 0 didn't help.

Am 07.01.23 um 11:17 schrieb Paul Kudla:

ok a little bit more info ....

My servers all run under FreeBSD 12.xx

which was also the base for the apple operating system origionally.

setting default_vsz_limit = 0 i seem to remember trying with no so great results

setting to zero can cause memory over runs (espically with replication) etc i found that when i used the config i sent eariler (vsz_limit is defaulted (not set )) everything worked

I generally find that tweaking the memory alloted to the individual services a more balanced approach but it does take a lot of trial and error

also note memory in the system is also a factor my mail servers have 32G dedicated to them which is what the settings were based on that seem to work pretty good at the moment.

I am running without any setting thus the default

I got this info from :

https://doc.dovecot.org/configuration_manual/service_configuration/

vsz_limit

Limit the process’s address space (both RLIMIT_DATA and RLIMIT_AS if available). When the space is reached, some memory allocations may start failing with “Out of memory”, or the kernel may kill the process with signal 9. This setting is mainly intended to prevent memory leaks from eating up all of the memory, but there can be also legitimate reasons why the process reaches this limit. For example a huge mailbox may not be accessed if this limit is too low. The default value (18446744073709551615=2^64-1) sets the limit to default_vsz_limit, while 0 disables the limit entirely.

There are 3 types of services that need to be optimized in different ways:

    Master services (e.g. auth, anvil, indexer, director, log):

        Currently there isn’t any easy way to optimize these. If these become a bottleneck, typically you need to run another Dovecot server. In some cases it may be possible to create multiple master processes and have each one be responsible for only specific users/processes, although this may also require some extra development.

    Services that do disk I/O or other blocking operations (e.g. imap, pop3, lmtp):

        These should have client_limit=1, because any blocking operation will block all the other clients and cause unnecessary delays and even timeouts. This means that process_limit specifies the maximum number of available parallel connections.

    Services that have no blocking operations (e.g. imap-login, pop3-login):

        For best performance (but a bit less safety), these should have process_limit and process_min_avail set to the number of CPU cores, so each CPU will be busy serving the process but without unnecessary context switches. Then client_limit needs to be set high enough to be able to serve all the needed connections (max connections=process_limit * client_limit). service_count is commonly set to unlimited (0) for these services. Otherwise when the service_count is beginning to be reached, the total number of available connections will shrink. With very bad luck that could mean that all the processes are simply waiting for the existing connections to die away before the process can die and a new one can be created. Although this could be made less likely by setting process_limit higher than process_min_avail, but that’s still not a guarantee since each process could get a very long running connection and the process_limit would be eventually reached.

Happy Saturday !!! Thanks - paul

Paul Kudla

Scom.ca Internet Services http://www.scom.ca 004-1009 Byron Street South Whitby, Ontario - Canada L1N 4S3

Toronto 416.642.7266 Main 1.866.411.7266 Fax 1.888.892.7266 Email paul@scom.ca

On 1/6/2023 5:20 PM, Gerben Wierda wrote:

...

How problematic is it to have

default_vsz_limit = 0

in dovecot.conf? macOS+MacPorts had this as a requirement even.

Gerben

...

On 6 Jan 2023, at 16:49, Paul Kudla mailto:paul@scom.ca> wrote:

i ran into this as well

here is the full config for mine with replication

# cat dovecot.conf # 2.3.14 (cee3cbc0d): /usr/local/etc/dovecot/dovecot.conf # OS: FreeBSD 12.1-RELEASE amd64 # Hostname: mail18.scom.ca http://mail18.scom.ca

auth_debug = no auth_debug_passwords = no

default_process_limit = 16384

mail_debug = no

#lock_method = dotlock #mail_max_lock_timeout = 300s

#mbox_read_locks = dotlock #mbox_write_locks = dotlock

mmap_disable = yes dotlock_use_excl = no mail_fsync = always mail_nfs_storage = no mail_nfs_index = no

auth_mechanisms = plain login auth_verbose = yes base_dir = /data/dovecot/run/ debug_log_path = syslog disable_plaintext_auth = no dsync_features = empty-header-workaround

info_log_path = syslog login_greeting = SCOM.CA Internet Services Inc. - Dovecot ready login_log_format_elements = user=<%u> method=%m rip=%r lip=%l mpid=%e %c

mail_location = maildir:~/

mail_plugins = " virtual notify replication fts fts_lucene " mail_prefetch_count = 20

protocols = imap pop3 lmtp sieve

protocol lmtp { mail_plugins = $mail_plugins sieve postmaster_address = }

service lmtp { process_limit=1000 vsz_limit = 512m client_limit=1 unix_listener /usr/home/postfix.local/private/dovecot-lmtp { group = postfix mode = 0600 user = postfix } }

protocol lda { mail_plugins = $mail_plugins sieve }

service lda { process_limit=1000 vsz_limit = 512m }

service imap { process_limit=4096 vsz_limit = 2g client_limit=1 }

service pop3 { process_limit=1000 vsz_limit = 512m client_limit=1 }

namespace inbox { inbox = yes location = mailbox Drafts { auto = subscribe special_use = \Drafts } mailbox Sent { auto = subscribe special_use = \Sent } mailbox Trash { auto = subscribe special_use = \Trash } prefix = separator = / }

passdb { args = /usr/local/etc/dovecot/dovecot-pgsql.conf driver = sql }

doveadm_port = 12345 doveadm_password = secretxxxx

service doveadm { process_limit = 0 process_min_avail = 0 idle_kill = 0 client_limit = 1 user = vmail inet_listener { port = 12345 } }

service config { unix_listener config { user = vmail } }

dsync_remote_cmd = ssh -l%{login} %{host} doveadm dsync-server -u%u #dsync_remote_cmd = doveadm sync -d -u%u

replication_dsync_parameters = -d -N -l 300 -U

plugin { mail_log_events = delete undelete expunge copy mailbox_delete mailbox_rename mail_log_fields = uid, box, msgid, from, subject, size, vsize, flags push_notification_driver = dlog

 sieve = file:~/sieve;active=~/sieve/.dovecot.sieve #sieve = ~/.dovecot.sieve sieve_duplicate_default_period = 1h sieve_duplicate_max_period = 1h sieve_extensions = +duplicate +notify +imapflags +vacation-seconds sieve_global_dir = /usr/local/etc/dovecot/sieve sieve_before = /usr/local/etc/dovecot/sieve/duplicates.sieve

 mail_replica = tcp:10.221.0.19:12345 #mail_replica = remote:vmail@10.221.0.19 mailto:vmail@10.221.0.19 #replication_sync_timeout = 2

 fts = lucene fts_lucene = whitespace_chars=@. fts_autoindex = yes fts_languages = en }

#sieve_extensions = vnd.dovecot.duplicate

#sieve_plugins = vnd.dovecot.duplicate

service anvil { process_limit = 1 client_limit=5000 vsz_limit = 512m unix_listener anvil { group = vmail mode = 0666 } }

service indexer-worker { vsz_limit = 2g }

service auth { process_limit = 1 client_limit=5000 vsz_limit = 1g

  unix_listener auth-userdb { mode = 0660 user = vmail group = vmail } unix_listener /var/spool/postfix/private/auth { mode = 0666 }

}

service stats { process_limit = 1000 vsz_limit = 1g unix_listener stats-reader { group = vmail mode = 0666 } unix_listener stats-writer { group = vmail mode = 0666 } } userdb { args = /usr/local/etc/dovecot/dovecot-pgsql.conf driver = sql

}

protocol imap { mail_max_userip_connections = 50 mail_plugins = $mail_plugins notify replication }

protocol pop3 { mail_max_userip_connections = 50 mail_plugins = $mail_plugins notify replication }

protocol imaps { mail_max_userip_connections = 25 mail_plugins = $mail_plugins notify replication }

protocol pop3s { mail_max_userip_connections = 25 mail_plugins = $mail_plugins notify replication }

service managesieve-login { process_limit = 1000 vsz_limit = 1g inet_listener sieve { port = 4190 } }

verbose_proctitle = yes

replication_max_conns = 100

replication_full_sync_interval = 1d

service replicator { client_limit = 0 drop_priv_before_exec = no idle_kill = 4294967295s process_limit = 1 process_min_avail = 0 service_count = 0 vsz_limit = 8g unix_listener replicator-doveadm { mode = 0600 user = vmail } vsz_limit = 8192M }

service aggregator { process_limit = 1000 #vsz_limit = 1g fifo_listener replication-notify-fifo { user = vmail group = vmail mode = 0666 }

}

service pop3-login { process_limit = 1000 client_limit = 100 vsz_limit = 512m }

service imap-urlauth-login { process_limit = 1000 client_limit = 1000 vsz_limit = 1g }

service imap-login { process_limit=1000 client_limit = 1000 vsz_limit = 1g }

protocol sieve { managesieve_implementation_string = Dovecot Pigeonhole managesieve_max_line_length = 65536 }

#Addition ssl config !include sni.conf

with sni cert support (examples)

# cat sni.conf #sni.conf ssl = yes verbose_ssl = yes ssl_dh =

#Default *.scom.ca http://scom.ca ssl_key =

local_name .scom.ca http://scom.ca { ssl_key =

}

local_name mail.clancyca.com http://mail.clancyca.com { ssl_key =http://mail.clancyca.com ssl_cert =http://mail.clancyca.com ssl_ca =http://mail.clancyca.com }

local_name mail.paulkudla.net http://mail.paulkudla.net { ssl_key =http://mail.paulkudla.net ssl_cert =http://mail.paulkudla.net ssl_ca =http://mail.paulkudla.net }

local_name mail.ekst.ca http://mail.ekst.ca { ssl_key =http://mail.ekst.ca ssl_cert =http://mail.ekst.ca ssl_ca =http://mail.ekst.ca }

local_name mail.hamletdevelopments.ca http://mail.hamletdevelopments.ca { ssl_key =http://mail.hamletdevelopments.ca ssl_cert =http://mail.hamletdevelopments.ca ssl_ca =http://mail.hamletdevelopments.ca }

pg sql support supporting replication

# cat dovecot-pgsql.conf driver = pgsql connect = host=localhost port=5433 dbname=scom_billing user=pgsql password= default_pass_scheme = PLAIN

password_query = SELECT username as user, password FROM email_users WHERE username = '%u' and password <> 'alias' and status = True and destination = '%u'

user_query = SELECT home, uid, gid FROM email_users WHERE username = '%u' and password <> 'alias' and status = True and destination = '%u'

#iterate_query = SELECT user, password FROM email_users WHERE username = '%u' and password <> 'alias' and status = True and destination = '%u'

iterate_query = SELECT "username" as user, domain FROM email_users WHERE status = True and alias_flag = False

Happy Friday !!! Thanks - paul

Paul Kudla

Scom.ca http://Scom.ca Internet Services <http://www.scom.ca http://www.scom.ca> 004-1009 Byron Street South Whitby, Ontario - Canada L1N 4S3

Toronto 416.642.7266 Main 1.866.411.7266 Fax 1.888.892.7266 Email paul@scom.ca mailto:paul@scom.ca

On 1/6/2023 5:32 AM, Gerben Wierda wrote:

...

...

On 6 Jan 2023, at 08:53, Aki Tuomi mailto:aki.tuomi@open-xchange.com> wrote:

On January 6, 2023 3:56:39 AM GMT+02:00, Gerben Wierda mailto:gerben.wierda@rna.nl> wrote:

...

One step further in my quest to create a replacement mail server.

I now have my old mail server (2.3.19.1, macOS + MacPorts) and my new (2.3.20, Alpine Linux, Docker, apk package). When I turn on replication it works, but, after a while I see:

Jan 06 00:50:31 replicator: Panic: data stack: Out of memory when allocating 268435496 bytes Jan 06 00:50:32 replicator: Fatal: master: service(replicator): child 133 killed with signal 6 (core dumped) Jan 06 00:50:32 lmtp(pid 195 user sysbh): Warning: replication(sysbh): Sync failure: Jan 06 00:50:32 lmtp(pid 195 user sysbh): Warning: replication(sysbh): Remote sent invalid input: -

I've removed synchronous operation for now (found a message on the net suggesting that) but is this known and what does it mean?

Gerben Wierda (LinkedIn <https://www.linkedin.com/in/gerbenwierda https://www.linkedin.com/in/gerbenwierda>) R&A IT Strategy <https://ea.rna.nl/ https://ea.rna.nl/> (main site) Book: Chess and the Art of Enterprise Architecture <https://ea.rna.nl/the-book/ https://ea.rna.nl/the-book/> Book: Mastering ArchiMate <https://ea.rna.nl/the-book-edition-iii/ https://ea.rna.nl/the-book-edition-iii/>

Dovecot default memory limit is 256M. You should probably set

service replicator { vsz_limit = 2G }

because replicator might have to use more memory, especially for larger indexes.

Aki That is a good tip as well. I had followed this bit of experience from someone else: https://marc.info/?l=dovecot&m=164438199727640 https://marc.info/?l=dovecot&m=164438199727640, haven't seen any err message since. But that might be because they are in sync now and both sides are aware. Can I trigger full replication again so I can test? Gerben

This message has been scanned for viruses and dangerous content by *MailScanner* http://www.mailscanner.info/, and is believed to be clean.

-- This message has been scanned for viruses and dangerous content by *MailScanner* http://www.mailscanner.info/, and is believed to be clean.