summaryrefslogtreecommitdiffstats
path: root/src/vnet/tcp/tcp.c
diff options
context:
space:
mode:
authorliuyacan <liuyacan@corp.netease.com>2021-06-14 18:09:01 +0800
committerFlorin Coras <florin.coras@gmail.com>2021-06-15 05:27:32 +0000
commit7e78119c257579731c8902556b4a197c3fc3e92e (patch)
tree2b8d7512b16aeb4ee76bfa3c0f47bf8a8a61afe7 /src/vnet/tcp/tcp.c
parent4d37bf9821d469f2c96ec3b2ddf8940320d9aa69 (diff)
tcp: prevent timer handler being called frequently
In the current implement, tcp would start or up an one tick retransmit timer for that connection if vlib_buffer_alloc() return 0. Now the tick is 0.1ms, this means that if VPP is in a buffer shortage state, there would be a large number of burst timer expirations. This commit limits the minimum interval of the retransmission timer to 100ms. Type: fix Signed-off-by: liuyacan <liuyacan@corp.netease.com> Change-Id: Ia11d693fe46119c5dc16b24ca93c30c31109057a
Diffstat (limited to 'src/vnet/tcp/tcp.c')
-rw-r--r--src/vnet/tcp/tcp.c1
1 files changed, 1 insertions, 0 deletions
diff --git a/src/vnet/tcp/tcp.c b/src/vnet/tcp/tcp.c
index 0ceac828d2d..c70a44cae90 100644
--- a/src/vnet/tcp/tcp.c
+++ b/src/vnet/tcp/tcp.c
@@ -1590,6 +1590,7 @@ tcp_configuration_init (void)
tcp_cfg.lastack_time = 300000; /* 30s */
tcp_cfg.finwait2_time = 300000; /* 30s */
tcp_cfg.closing_time = 300000; /* 30s */
+ tcp_cfg.alloc_err_timeout = 1000; /* 100ms */
/* This value is seconds */
tcp_cfg.cleanup_time = 0.1; /* 100ms */