NetworkDB to honor the Network Control Plane MTU #1839
Conversation
|
This is the preliminary step, still needs the value to be set, so far will keep using the default one of 1400 |
fcrisciani
force-pushed
the
datapath-mtu
branch
8 times, most recently
from
ae47a6c
to
d61fe86
Compare
fcrisciani
changed the title
fcrisciani
force-pushed
the
datapath-mtu
branch
4 times, most recently
from
d4877b3
to
9bf8207
Compare
config/config.go
Outdated
| if exp < 1500 { | ||
| // if exp == 0 the value won't be used | ||
| logrus.Warnf("Received a Network Control Plane MTU of %d, this value is very LOW,", | ||
| "the database can misbehave, 0 value will be ignored", exp) |
There was a problem hiding this comment.
LOW -> low pls :)
database -> network control-plane
pls remove 0 value will be ignored ... I don't think it is required.
There was a problem hiding this comment.
ok for low also if I wanted to convey fear :D
ok for network control-plane
I put the 0 value because here https://github.com/docker/libnetwork/pull/1839/files#diff-2fe34e1b6f7f9aece0af74b1635502eeR217 it is possible to see 0 if the user set it so.
networkdb/cluster.go
Outdated
| if printStats { | ||
| healthScore := nDB.memberlist.GetHealthScore() | ||
| if healthScore == 0 { | ||
| logrus.Infof("NetworkDB healthscore:%d Node totally healthy", healthScore) |
There was a problem hiding this comment.
Is this going to be printed every 5 seconds ? I think that is very aggressive.
There was a problem hiding this comment.
The definition of totally healthy comes directly from memberlist:
GetHealthScore gives this instance's idea of how well it is meeting the soft real-time requirements of the protocol. Lower numbers are better, and zero means "totally healthy".
Will change
|
@fcrisciani minor comments. Otherwise LGTM |
fcrisciani
force-pushed
the
datapath-mtu
branch
3 times, most recently
from
1f554b8
to
7f3a0e6
Compare
- Introduce the possibility to specify the max buffer length in network DB. This will allow to use the whole MTU limit of the interface - Add queue stats per network, it can be handy to identify the node's throughput per network and identify unbalance between nodes that can point to an MTU missconfiguration Signed-off-by: Flavio Crisciani <flavio.crisciani@docker.com>
Introduce the possibility to specify the max buffer length
in network DB. This will allow to pass the proper MTU to
networkDB
Signed-off-by: Flavio Crisciani flavio.crisciani@docker.com