Article ID: 126063, created on Jul 1, 2015, last review on Sep 22, 2015

  • Applies to:
  • Virtuozzo 6.0
  • Virtuozzo containers for Linux


Starting Virtuozzo services (or a virtual environment) on a server with connected Virtuozzo storage, the process can stuck without any output or error message printed.

At the same time, commands like pstorage -c ${CLUSTER_NAME} stat or directory listing of mounted Virtuozzo storage share work without any issue.

Read operations with any file stored in Virtuozzo storage hangs too, resulting in hanging Virtuozzo-related services.


Starting vz service hangs right after mounting pfcache:

~# /etc/init.d/vz start
Starting Parallels Cloud Server:                           [  OK  ]
Load OOM groups                                            [  OK  ]
Starting vzeventd:                                         [  OK  ]
Could not change any device features
Could not change any device features
Loading Parallels Cloud Server license:                    [  OK  ]
Set vzprivrange:                                           [  OK  ]
Configuring Parallels Cloud Server accounting:             [  OK  ]
Configuring Parallels Cloud Server shaping:                [  OK  ]
Waiting for license monitor start:.
Mount pfcache image /vz/pfcache.hdd                        [  OK  ]

Stack of the process reading a file might look like this:

~# cat /proc/59516/stack
[<ffffffff8113e16d>] sync_page+0x3d/0x50
[<ffffffff8113e18e>] sync_page_killable+0xe/0x40
[<ffffffff8113e097>] __lock_page_killable+0x67/0x70
[<ffffffff8114011d>] generic_file_read_iter+0x3cd/0x640
[<ffffffff8114041b>] generic_file_aio_read+0x8b/0xa0
[<ffffffffa037c34b>] fuse_file_aio_read+0x7b/0xb0 [fuse]
[<ffffffff811b3e1a>] do_sync_read+0xfa/0x140
[<ffffffff811b47b5>] vfs_read+0xb5/0x1a0
[<ffffffff811b48f1>] sys_read+0x51/0x90
[<ffffffff8100b142>] system_call_fastpath+0x16/0x1b
[<ffffffffffffffff>] 0xffffffffffffffff

The log file/var/log/pstorage/ contains records showing 5 seconds timeout on processing a request:

19-06-15 20:06:04.087 pcs_map_complete: [2.604808:13761] received map 36:6:3:1:37 for 0@[0002406f]/ve.conf -> O00028b79
19-06-15 20:06:04.087 cs_whitelist: Whitelisting CS1029 by mds hint
19-06-15 20:06:04.087 pcs_map_complete: 0@[0002406f]/ve.conf -> O00028b79 1 node map { 1029 0 0}
19-06-15 20:06:04.088 pcs_cc_process_ireq_chunk: [0002406f]/ve.conf [0]
19-06-15 20:06:04.088 pcs_flow_cs_analysis: CS Flow age 172049, 0+1924
19-06-15 20:06:04.088 log_cs_info: LAT CS#1029 io_prio=68[0] lat=0+0/0 infl=24664/0/1 p=0 csf=1/0(1/1)
19-06-15 20:06:04.088 pcs_cslist_submit_read: Selected read map 0@[0002406f]/ve.conf to CS1029; is_seq=0
19-06-15 20:06:11.181 kill_timer_work: killing msg to CS#1029 type=256 xid=[2.604808:13762] stage=1 tmo=6000 exp=-1425069515062 rem=-1425069491062
19-06-15 20:06:11.181 cs_response_done: [2.604808:13762]IO error 14 1029
19-06-15 20:06:11.181 cs_blacklist: Blacklisting CS1029 by read error, err=14

Firewall rules are too restrictive, with default policy for INPUT chain set to DROP.


Since listing files in mounted Virtuozzo storage works, communication in Virtuozzo cluster for metadata information does not have any issue.

What does not work properly is connecting to chunk servers, read operations from the local and remote chunk servers are not allowed, packets are dropped.


Reconfigure firewall to allow connections through the interface which is dedicated for Virtuozzo storage.

Search Words

vzlist didn't return list of containers

vz hungs on start pcss

c62e8726973f80975db0531f1ed5c6a2 2897d76d56d2010f4e3a28f864d69223 0dd5b9380c7d4884d77587f3eb0fa8ef e8e50b42231236b82df27684e7ec0beb d02f9caf3e11b191a38179103495106f

Email subscription for changes to this article
Save as PDF