On 4/14/20 5:57 AM, Tuong Tong Lien wrote:
Hi Jon,
As mentioned in the last meeting, we got a report from customer that TIPC has
some latency increased (~ 30-50%) in message streaming, it's after backporting
some recent patches. By only disabling/enabling the Nagle feature, we are now
able
On 4/14/20 3:30 AM, Tuong Lien wrote:
Since commit 6a939f365bdb ("tipc: Auto removal of peer down node
instance"), a peer node instance is automatically clean up after 5 mins
without contact. This changed behavior is not really expected from user
who would expect to see the node with the
Hmm, I was sure I already did.
But anyway:
Acked-by: Jon Maloy
On 4/14/20 3:54 AM, Tuong Tong Lien wrote:
Hi Jon, all,
I understand that you have agreed on the solution to the issue in previous discussions,
but please make a formal "ack-by" so I will send it to net.
Thanks a lot!
BR/Tuong
Hi Jon,
As mentioned in the last meeting, we got a report from customer that TIPC has
some latency increased (~ 30-50%) in message streaming, it's after backporting
some recent patches. By only disabling/enabling the Nagle feature, we are now
able to confirm this Nagle patch causing the
Hi a.Hoang,
I have really tried to make it as "short" as possible because of the
"80-character" rule. Moreover, the value is not only a time value but also
playing as a "switch" to turn off/on the auto removal feature (see the commit
message). Anyway, it is just a name so we can change it to
Hi Tuong,
The 'node_cleanup' is not clear in meaning. I'd prefer using
'node_cleanup_time_secs'.
I think we should use 'milliseconds' for more popular than 'seconds' in
proc/sysctl.
Regards,
Hoang
-Original Message-
From: Tuong Tong Lien
Sent: Tuesday, April 14, 2020 2:30 PM
To:
Hi Jon, all,
I understand that you have agreed on the solution to the issue in previous
discussions, but please make a formal "ack-by" so I will send it to net.
Thanks a lot!
BR/Tuong
-Original Message-
From: Tuong Lien
Sent: Friday, March 27, 2020 7:31 PM
To: jma...@redhat.com;
Since commit 6a939f365bdb ("tipc: Auto removal of peer down node
instance"), a peer node instance is automatically clean up after 5 mins
without contact. This changed behavior is not really expected from user
who would expect to see the node with the "down" state in the node list
as before