github/kubernetes
1
0
Fork 0
mirror of https://github.com/k3s-io/kubernetes.git synced 2025-07-23 19:56:01 +00:00
Code Issues Projects Releases Wiki Activity

Don't re-run EnsureChain/EnsureRules on partial syncs

Browse Source 
We currently invoke /sbin/iptables 24 times on each syncProxyRules
before calling iptables-restore. Since even trivial iptables
invocations are slow on hosts with lots of iptables rules, this adds a
lot of time to each sync. Since these checks are expected to be a
no-op 99% of the time, skip them on partial syncs.
This commit is contained in:
Dan Winship 2022-11-29 09:38:40 -05:00
parent 7061704297
commit bfa4948bb6
1 changed files with 25 additions and 13 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
pkg/proxy/iptables/proxier.go
View File

@ -863,7 +863,18 @@ func (proxier *Proxier) syncProxyRules() {
}
}()
// Create and link the kube chains.
if !tryPartialSync {
// Ensure that our jump rules (eg from PREROUTING to KUBE-SERVICES) exist.
// We can't do this as part of the iptables-restore because we don't want
// to specify/replace *all* of the rules in PREROUTING, etc.
//
// We need to create these rules when kube-proxy first starts, and we need
// to recreate them if the utiliptables Monitor detects that iptables has
// been flushed. In both of those cases, the code will force a full sync.
// In all other cases, it ought to be safe to assume that the rules
// already exist, so we'll skip this step when doing a partial sync, to
// save us from having to invoke /sbin/iptables 20 times on each sync
// (which will be very slow on hosts with lots of iptables rules).
for _, jump := range iptablesJumpChains {
if _, err := proxier.iptables.EnsureChain(jump.table, jump.dstChain); err != nil {
klog.ErrorS(err, "Failed to ensure chain exists", "table", jump.table, "chain", jump.dstChain)
@ -878,6 +889,7 @@ func (proxier *Proxier) syncProxyRules() {
return
}
}
}
//
// Below this point we will not return until we try to write the iptables rules.