Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

zebra: reduce impact of route-update overload #6853

Merged
merged 2 commits into from
Jan 13, 2021
Merged

zebra: reduce impact of route-update overload #6853

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
0ca6f3b
zebra: remove useless deleted route_entries promptly
Jul 22, 2020
a88a7c8
zebra: improve dataplane plugin queue counters
Jul 28, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Next Next commit
zebra: remove useless deleted route_entries promptly 
Zebra accumulates route-entry objects and then processes them
as a group. If that rib processing is delayed, because the
dataplane/fib programming has built up a queue e.g., zebra can
hold multiple deleted route objects in memory. At scale, this can
be a problem. Delete unneeded route entries promptly, if they
can't contribute to rib processing.

Signed-off-by: Mark Stapp <mjs@voltanet.io>
  • Loading branch information
Mark Stapp committed Dec 7, 2020
commit 0ca6f3b1e69529f5fe924ff19860d5b958d6e4a8
41 changes: 35 additions & 6 deletions zebra/zebra_rib.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2933,8 +2933,10 @@ int rib_add_multipath_nhe(afi_t afi, safi_t safi, struct prefix *p,
struct nhg_hash_entry *nhe = NULL;
struct route_table *table;
struct route_node *rn;
struct route_entry *same = NULL;
struct route_entry *same = NULL, *first_same = NULL;
int ret = 0;
int same_count = 0;
rib_dest_t *dest;

if (!re || !re_nhe)
return -1;
Expand Down Expand Up @@ -3002,14 +3004,22 @@ int rib_add_multipath_nhe(afi_t afi, safi_t safi, struct prefix *p,
* for the install don't do a route replace.
*/
RNODE_FOREACH_RE (rn, same) {
if (CHECK_FLAG(same->status, ROUTE_ENTRY_REMOVED))
if (CHECK_FLAG(same->status, ROUTE_ENTRY_REMOVED)) {
same_count++;
continue;
}

/* Compare various route_entry properties */
if (rib_compare_routes(re, same))
break;
if (rib_compare_routes(re, same)) {
same_count++;

if (first_same == NULL)
first_same = same;
}
}

same = first_same;

/* If this route is kernel/connected route, notify the dataplane. */
if (RIB_SYSTEM_ROUTE(re)) {
/* Notify dataplane */
Expand All @@ -3019,8 +3029,9 @@ int rib_add_multipath_nhe(afi_t afi, safi_t safi, struct prefix *p,
/* Link new re to node.*/
if (IS_ZEBRA_DEBUG_RIB) {
rnode_debug(rn, re->vrf_id,
"Inserting route rn %p, re %p (%s) existing %p",
rn, re, zebra_route_string(re->type), same);
"Inserting route rn %p, re %p (%s) existing %p, same_count %d",
rn, re, zebra_route_string(re->type), same,
same_count);

if (IS_ZEBRA_DEBUG_RIB_DETAILED)
route_entry_dump(p, src_p, re);
Expand All @@ -3034,6 +3045,24 @@ int rib_add_multipath_nhe(afi_t afi, safi_t safi, struct prefix *p,
if (same)
rib_delnode(rn, same);

/* See if we can remove some RE entries that are queued for
* removal, but won't be considered in rib processing.
*/
dest = rib_dest_from_rnode(rn);
RNODE_FOREACH_RE_SAFE (rn, re, same) {
if (CHECK_FLAG(re->status, ROUTE_ENTRY_REMOVED)) {
/* If the route was used earlier, must retain it. */
if (dest && re == dest->selected_fib)
continue;

if (IS_ZEBRA_DEBUG_RIB)
rnode_debug(rn, re->vrf_id, "rn %p, removing unneeded re %p",
rn, re);

rib_unlink(rn, re);
}
}

route_unlock_node(rn);
return ret;
}
Expand Down