mirrors/scoutfs
1
0
Fork 0
mirror of https://github.com/versity/scoutfs.git synced 2026-09-26 18:04:40 +00:00
Code Issues Packages Projects Releases Wiki Activity

Add tcp_keepalive_timeout_ms option.

Browse Source 
The default TCP keepalive value is currently 10s, resulting in clients
being disconnected after 10 seconds of not replying to a TCP keepalive
packet. These keepalive values are reasonable most of the times, but
we've seen client disconnects where this timeout has been exceeded,
resulting in fencing. The cause for this is unknown at this time, but it
is suspected that network intermissions are happening.

This change adds a configurable value for this specific client socket
timeout. It enforces that its value is above UNRESPONSIVE_PROBES, whose
value remains unchanged.

The default value of 10000ms (10s) remains the trusted value. It is
enirely unclear and untested what values are reasonable and which
ones are not.  Since the value of this setting can and will interact
with other timeout values, care must be taken to not exceed certain
other timeout values.  I've tested this only briefly with values of
5000 and 25000. Outside that range is likely problematic.

Signed-off-by: Auke Kok <auke.kok@versity.com>
This commit is contained in:
Auke Kok
2025-10-07 12:16:23 -07:00
parent 65ea250de9
commit ad79ee94f9
4 changed files with 61 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files
utils/man/scoutfs.5
+18
View File
@@ -130,6 +130,24 @@ the server for the filesystem if it is elected leader.
The assigned number must match one of the slots defined with \-Q options
when the filesystem was created with mkfs. If the number assigned
doesn't match a number created during mkfs then the mount will fail.
.TP
.B tcp_keepalive_timeout_ms=<number>
This option sets the amount of time, in milliseconds, that a client
connection will wait for active TCP packets, before deciding that
the connection is dead. This setting is per-mount and only changes
the behavior of that mount.
.sp
The default value of this setting is 10000msec (10s). Any precision
beyond a whole second is likely unrealistic due to the nature of
TCP keepalive mechanisms in the Linux kernel. Valid values are any
value higher than 3000 (3s). Values that are higher than 30000msec
(30s) will likely interfere with other embedded timeout values.
.sp
The TCP keepalive mechanism is complex and observing a lost connection
quickly is important to maintain cluster stability. If the local
network suffers from intermittent outages this option may provide
some respite to overcome these outages without the cluster becoming
desynchronized.
.SH VOLUME OPTIONS
Volume options are persistent options which are stored in the super
block in the metadata device and which apply to all mounts of the volume.