From: Daniel Kral <d.kral@proxmox.com>
To: pve-devel@lists.proxmox.com
Subject: [PATCH ha-manager v4 14/17] lrm: read service config once per work iteration
Date: Wed, 18 Nov 2026 15:44:59 +0100 [thread overview]
Message-ID: <20261118144502.231134-15-d.kral@proxmox.com> (raw)
In-Reply-To: <20261118144502.231134-1-d.kral@proxmox.com>
The amount of times the LRM reads and parses the HA resource config
primarily scales with the amount of HA resource command exit codes that
need to be handled. This can become quite expensive especially with many
HA resources in 'started' state, where the HA Manager issues an HA
resource command on every HA Manager round.
In the PVE2 environment, the HA resource config and vmlist is cached
between PVE::Cluster::cfs_update() calls, and in the Test environment
Sim Hardware commands can only change the HA resource config outside of
the LRM round window.
Therefore, the return value of read_service_config() can be cached
inbetween and must only be invalidated again once a new LRM round
starts.
This significantly reduces the time spent reading and parsing the HA
resource config every time run_workers() is called and every time a LRM
worker's result is collected with handle_service_exitcode().
A test startup of 768 HA resources on a single node at the same time
showed ~99.2% fewer calls to checked_resources_config() (from 10,333 to
82 calls) and ~99.3% fewer calls to parse_sid() (from 15,639,537 to
109,198 calls).
Additionally, this reduces the inclusive average execution time per call
for several subroutines where the HA resource config was read within:
+---------------------------+---------------+--------------+
| Subroutine | Before Change | After Change |
+---------------------------+---------------+--------------+
| check_active_workers | 327ms | 11.6ms |
| resource_command_finished | 22.7ms | 2.81ms |
| handle_service_exitcode | 21.1ms | 10µs |
+---------------------------+---------------+--------------+
The profiling was done on a physical host with Devel::NYTProf 6.14 [0].
[0] https://metacpan.org/pod/Devel::NYTProf
Signed-off-by: Daniel Kral <d.kral@proxmox.com>
Reviewed-by: Michael Köppl <m.koeppl@proxmox.com>
Tested-by: Michael Köppl <m.koeppl@proxmox.com>
---
src/PVE/HA/LRM.pm | 25 +++++++++++++++++++++++--
1 file changed, 23 insertions(+), 2 deletions(-)
diff --git a/src/PVE/HA/LRM.pm b/src/PVE/HA/LRM.pm
index ab78d274..441404d2 100644
--- a/src/PVE/HA/LRM.pm
+++ b/src/PVE/HA/LRM.pm
@@ -41,6 +41,9 @@ sub new {
mode => 'active',
cluster_state_update => 0,
active_idle_rounds => 0,
+ # service config cache in current work() iteration
+ # does contain all HA resources, users must filter for the assigned node
+ service_config => undef,
# service_status from manager_status in current work() iteration
# does contain all HA resources, users must filter for the assigned node
service_status => {},
@@ -247,6 +250,22 @@ sub update_lrm_status {
}
}
+sub read_service_config {
+ my ($self) = @_;
+
+ return $self->{service_config} if defined($self->{service_config});
+
+ $self->{service_config} = $self->{haenv}->read_service_config();
+
+ return $self->{service_config};
+}
+
+sub invalidate_service_config_cache {
+ my ($self) = @_;
+
+ undef $self->{service_config};
+}
+
sub get_protected_ha_agent_lock {
my ($self) = @_;
@@ -349,6 +368,8 @@ sub do_one_iteration {
$self->{cluster_state_update} = $haenv->cluster_state_update();
+ $self->invalidate_service_config_cache();
+
my $res = $self->work();
$haenv->loop_end_hook();
@@ -668,7 +689,7 @@ sub run_workers {
# number of workers to start, if 0 we exec the command directly without forking
my $max_workers = $haenv->get_max_workers();
- my $sc = $haenv->read_service_config();
+ my $sc = $self->read_service_config();
my $worker = $self->{workers};
# we only got limited time but want to ensure that every queued worker is scheduled
@@ -916,7 +937,7 @@ sub handle_service_exitcode {
my $haenv = $self->{haenv};
my $tries = $self->{restart_tries};
- my $sc = $haenv->read_service_config();
+ my $sc = $self->read_service_config();
my $max_restart = 0;
--
2.47.3
next prev parent reply other threads:[~2026-11-18 14:47 UTC|newest]
Thread overview: 22+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-11-18 14:44 [PATCH-SERIES ha-manager v4 00/17] some sid parsing and LRM speedup improvements Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 01/17] resources: remove commented ipaddr resource type Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 02/17] tree-wide: use the term vmid instead of name when referring to VMs/CTs Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 03/17] tree-wide: remove unused return values at parse_sid callsites Daniel Kral
2026-11-18 15:20 ` Michael Köppl
2026-11-18 14:44 ` [PATCH ha-manager v4 04/17] config: make update_single_resource_config_inplace only allow sids Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 05/17] make parse_sid always return an array Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 06/17] config: use early returns in parse_sid Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 07/17] config: introduce separate parse_vmid_or_sid helper subroutine Daniel Kral
2026-11-18 16:02 ` Michael Köppl
2026-11-18 14:44 ` [PATCH ha-manager v4 08/17] drop the unmodified sid return array entry value from parse_sid Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 09/17] lrm: catch and report dies while the LRM is in maintenance state Daniel Kral
2026-11-18 16:12 ` Michael Köppl
2026-11-18 14:44 ` [PATCH ha-manager v4 10/17] lrm: rename update_lrm_status to flush_lrm_status Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 11/17] lrm: rename update_service_status to update_lrm_status Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 12/17] lrm: document and initialize LRM instance properties Daniel Kral
2026-11-18 14:44 ` [PATCH ha-manager v4 13/17] lrm: update the service status once per work iteration Daniel Kral
2026-11-18 14:44 ` Daniel Kral [this message]
2026-11-18 14:45 ` [PATCH ha-manager v4 15/17] lrm: compute valid service uids hash set " Daniel Kral
2026-11-18 14:45 ` [PATCH ha-manager v4 16/17] lrm: resource_command_finished: avoid writing stale LRM results Daniel Kral
2026-11-18 14:45 ` [PATCH ha-manager v4 17/17] lrm: prune stale LRM results once per work iteration Daniel Kral
2026-11-18 20:39 ` applied: [PATCH-SERIES ha-manager v4 00/17] some sid parsing and LRM speedup improvements Thomas Lamprecht
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261118144502.231134-15-d.kral@proxmox.com \
--to=d.kral@proxmox.com \
--cc=pve-devel@lists.proxmox.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox