Re: [PATCH v2] ipvs: Fix reuse connection if RS weight is 0
From: Julian Anastasov <ja@ssi.bg>
Date: 2021-10-29 19:25:41
Also in:
lkml, lvs-devel, netdev, netfilter-devel
Hello, On Fri, 29 Oct 2021, yangxingwu wrote:
Since commit dc7b3eb900aa ("ipvs: Fix reuse connection if real server is
dead"), new connections to dead servers are redistributed immediately to
new servers.
Then commit d752c3645717 ("ipvs: allow rescheduling of new connections when
port reuse is detected") disable expire_nodest_conn if conn_reuse_mode is
0. And new connection may be distributed to a real server with weight 0.Can you better explain in commit message that we are changing expire_nodest_conn to work even for reused connections when conn_reuse_mode=0 but without affecting the controlled/persistent connections during the grace period while server is with weight=0. Even if you target -next trees adding commit d752c3645717 as Fixes line would be a good idea. Make sure the tree is specified after the v3 tag.
quoted hunk ↗ jump to hunk
Co-developed-by: Chuanqi Liu <redacted> Signed-off-by: Chuanqi Liu <redacted> Signed-off-by: yangxingwu <redacted> --- Documentation/networking/ipvs-sysctl.rst | 3 +-- net/netfilter/ipvs/ip_vs_core.c | 7 ++++--- 2 files changed, 5 insertions(+), 5 deletions(-)diff --git a/Documentation/networking/ipvs-sysctl.rst b/Documentation/networking/ipvs-sysctl.rst index 2afccc63856e..1cfbf1add2fc 100644 --- a/Documentation/networking/ipvs-sysctl.rst +++ b/Documentation/networking/ipvs-sysctl.rst@@ -37,8 +37,7 @@ conn_reuse_mode - INTEGER 0: disable any special handling on port reuse. The new connection will be delivered to the same real server that was - servicing the previous connection. This will effectively - disable expire_nodest_conn. + servicing the previous connection. bit 1: enable rescheduling of new connections when it is safe. That is, whenever expire_nodest_conn and for TCP sockets, whendiff --git a/net/netfilter/ipvs/ip_vs_core.c b/net/netfilter/ipvs/ip_vs_core.c index 128690c512df..374f4b0b7080 100644 --- a/net/netfilter/ipvs/ip_vs_core.c +++ b/net/netfilter/ipvs/ip_vs_core.c@@ -2042,14 +2042,15 @@ ip_vs_in(struct netns_ipvs *ipvs, unsigned int hooknum, struct sk_buff *skb, int ipvs, af, skb, &iph); conn_reuse_mode = sysctl_conn_reuse_mode(ipvs); - if (conn_reuse_mode && !iph.fragoffs && is_new_conn(skb, &iph) && cp) { + if (!iph.fragoffs && is_new_conn(skb, &iph) && cp) {
It is even better to move the !cp->control check above:
if (!iph.fragoffs && is_new_conn(skb, &iph) && cp && !cp->control) {
Then is not needed in is_new_conn_expected() anymore.
bool old_ct = false, resched = false;
And now you can move conn_reuse_mode here: int conn_reuse_mode = sysctl_conn_reuse_mode(ipvs);
if (unlikely(sysctl_expire_nodest_conn(ipvs)) && cp->dest &&
- unlikely(!atomic_read(&cp->dest->weight))) {
+ unlikely(!atomic_read(&cp->dest->weight)) && !cp->control) {
resched = true;
old_ct = ip_vs_conn_uses_old_conntrack(cp, skb);
- } else if (is_new_conn_expected(cp, conn_reuse_mode)) {
+ } else if (conn_reuse_mode &&
+ is_new_conn_expected(cp, conn_reuse_mode)) {
old_ct = ip_vs_conn_uses_old_conntrack(cp, skb);
if (!atomic_read(&cp->n_control)) {
resched = true;
--
2.30.2Regards -- Julian Anastasov [off-list ref]