mir/ovs
2
0
Fork 0
mirror of https://github.com/openvswitch/ovs synced 2025-08-22 09:58:01 +00:00
Code Issues Releases Wiki Activity

dpif-netdev: Add parameters to configure PMD auto load balance.

Browse Source 
Two important parts of how PMD auto load balance operates are how
loaded a core needs to be and how much improvement is estimated
before a PMD auto load balance can trigger.

Previously they were hardcoded to 95% loaded and 25% variance
improvement.

These default values may not be suitable for all use cases and
we may want to use a more (or less) aggressive rebalance, either
on the pmd load threshold or on the minimum variance improvement
threshold.

The defaults are not changed, but "pmd-auto-lb-load-threshold" and
"pmd-auto-lb-improvement-threshold" parameters are added to override
the defaults.

$ ovs-vsctl set open_vswitch . other_config:pmd-auto-lb-load-threshold="70"
$ ovs-vsctl set open_vswitch . other_config:pmd-auto-lb-improvement-threshold="20"

Signed-off-by: Christophe Fontaine <cfontain@redhat.com>
Co-Authored-by: Kevin Traynor <ktraynor@redhat.com>
Signed-off-by: Kevin Traynor <ktraynor@redhat.com>
Acked-by: David Marchand <david.marchand@redhat.com>
Acked-by: Ian Stokes <ian.stokes@intel.com>
Signed-off-by: Ilya Maximets <i.maximets@ovn.org>
This commit is contained in:
Christophe Fontaine 2020-12-21 13:10:31 +00:00 committed by Ilya Maximets
parent e4db0b69e2
commit 62ab5594c2
3 changed files with 79 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
NEWS
View File

@ -20,6 +20,9 @@ Post-v2.14.0
* Add generic IP protocol support to conntrack. With this change, all
none UDP, TCP, and ICMP traffic will be treated as general L3
traffic, i.e. using 3 tupples.
* Add parameters 'pmd-auto-lb-load-threshold' and
'pmd-auto-lb-improvement-threshold' to configure PMD auto load balance
behaviour.
- The environment variable OVS_UNBOUND_CONF, if set, is now used
as the DNS resolver's (unbound) configuration file.
- Linux datapath:

55
lib/dpif-netdev.c
View File

@ -85,9 +85,9 @@
VLOG_DEFINE_THIS_MODULE(dpif_netdev);
/* Auto Load Balancing Defaults */
#define ALB_ACCEPTABLE_IMPROVEMENT 25
#define ALB_PMD_LOAD_THRESHOLD 95
#define ALB_PMD_REBALANCE_POLL_INTERVAL 1 /* 1 Min */
#define ALB_IMPROVEMENT_THRESHOLD 25
#define ALB_LOAD_THRESHOLD 95
#define ALB_REBALANCE_INTERVAL 1 /* 1 Min */
#define MIN_TO_MSEC 60000
#define FLOW_DUMP_MAX_BATCH 50
@ -300,6 +300,8 @@ struct pmd_auto_lb {
bool is_enabled; /* Current status of Auto load balancing. */
uint64_t rebalance_intvl;
uint64_t rebalance_poll_timer;
uint8_t rebalance_improve_thresh;
atomic_uint8_t rebalance_load_thresh;
};
/* Datapath based on the network device interface from netdev.h.
@ -4204,6 +4206,7 @@ set_pmd_auto_lb(struct dp_netdev *dp)
unsigned int cnt = 0;
struct dp_netdev_pmd_thread *pmd;
struct pmd_auto_lb *pmd_alb = &dp->pmd_alb;
uint8_t rebalance_load_thresh;
bool enable_alb = false;
bool multi_rxq = false;
@ -4233,9 +4236,16 @@ set_pmd_auto_lb(struct dp_netdev *dp)
if (pmd_alb->is_enabled != enable_alb) {
pmd_alb->is_enabled = enable_alb;
if (pmd_alb->is_enabled) {
atomic_read_relaxed(&pmd_alb->rebalance_load_thresh,
&rebalance_load_thresh);
VLOG_INFO("PMD auto load balance is enabled "
"interval %"PRIu64" mins",
pmd_alb->rebalance_intvl / MIN_TO_MSEC);
"interval %"PRIu64" mins, "
"pmd load threshold %"PRIu8"%%, "
"improvement threshold %"PRIu8"%%",
pmd_alb->rebalance_intvl / MIN_TO_MSEC,
rebalance_load_thresh,
pmd_alb->rebalance_improve_thresh);
} else {
pmd_alb->rebalance_poll_timer = 0;
VLOG_INFO("PMD auto load balance is disabled");
@ -4259,6 +4269,8 @@ dpif_netdev_set_config(struct dpif *dpif, const struct smap *other_config)
uint32_t insert_min, cur_min;
uint32_t tx_flush_interval, cur_tx_flush_interval;
uint64_t rebalance_intvl;
uint8_t rebalance_load, cur_rebalance_load;
uint8_t rebalance_improve;
tx_flush_interval = smap_get_int(other_config, "tx-flush-interval",
DEFAULT_TX_FLUSH_INTERVAL);
@ -4336,7 +4348,7 @@ dpif_netdev_set_config(struct dpif *dpif, const struct smap *other_config)
false);
rebalance_intvl = smap_get_int(other_config, "pmd-auto-lb-rebal-interval",
ALB_PMD_REBALANCE_POLL_INTERVAL);
ALB_REBALANCE_INTERVAL);
/* Input is in min, convert it to msec. */
rebalance_intvl =
@ -4348,6 +4360,30 @@ dpif_netdev_set_config(struct dpif *dpif, const struct smap *other_config)
"%"PRIu64" mins\n", rebalance_intvl / MIN_TO_MSEC);
}
rebalance_improve = smap_get_int(other_config,
"pmd-auto-lb-improvement-threshold",
ALB_IMPROVEMENT_THRESHOLD);
if (rebalance_improve > 100) {
rebalance_improve = ALB_IMPROVEMENT_THRESHOLD;
}
if (rebalance_improve != pmd_alb->rebalance_improve_thresh) {
pmd_alb->rebalance_improve_thresh = rebalance_improve;
VLOG_INFO("PMD auto load balance improvement threshold set to "
"%"PRIu8"%%", rebalance_improve);
}
rebalance_load = smap_get_int(other_config, "pmd-auto-lb-load-threshold",
ALB_LOAD_THRESHOLD);
if (rebalance_load > 100) {
rebalance_load = ALB_LOAD_THRESHOLD;
}
atomic_read_relaxed(&pmd_alb->rebalance_load_thresh, &cur_rebalance_load);
if (rebalance_load != cur_rebalance_load) {
atomic_store_relaxed(&pmd_alb->rebalance_load_thresh,
rebalance_load);
VLOG_INFO("PMD auto load balance load threshold set to %"PRIu8"%%",
rebalance_load);
}
set_pmd_auto_lb(dp);
return 0;
}
@ -5676,7 +5712,7 @@ pmd_rebalance_dry_run(struct dp_netdev *dp)
improvement =
((curr_variance - new_variance) * 100) / curr_variance;
}
if (improvement < ALB_ACCEPTABLE_IMPROVEMENT) {
if (improvement < dp->pmd_alb.rebalance_improve_thresh) {
ret = false;
}
}
@ -8711,6 +8747,7 @@ dp_netdev_pmd_try_optimize(struct dp_netdev_pmd_thread *pmd,
if (pmd->ctx.now > pmd->rxq_next_cycle_store) {
uint64_t curr_tsc;
uint8_t rebalance_load_trigger;
struct pmd_auto_lb *pmd_alb = &pmd->dp->pmd_alb;
if (pmd_alb->is_enabled && !pmd->isolated
&& (pmd->perf_stats.counters.n[PMD_CYCLES_ITER_IDLE] >=
@ -8727,7 +8764,9 @@ dp_netdev_pmd_try_optimize(struct dp_netdev_pmd_thread *pmd,
pmd_load = ((tot_proc * 100) / (tot_idle + tot_proc));
}
if (pmd_load >= ALB_PMD_LOAD_THRESHOLD) {
atomic_read_relaxed(&pmd_alb->rebalance_load_thresh,
&rebalance_load_trigger);
if (pmd_load >= rebalance_load_trigger) {
atomic_count_inc(&pmd->pmd_overloaded);
} else {
atomic_count_set(&pmd->pmd_overloaded, 0);

31
vswitchd/vswitch.xml
View File

@ -653,8 +653,9 @@
type='{"type": "boolean"}'>
<p>
Configures PMD Auto Load Balancing that allows automatic assignment of
RX queues to PMDs if any of PMDs is overloaded (i.e. processing cycles
> 95%).
RX queues to PMDs if any of PMDs is overloaded (i.e. a processing
cycles >
<ref column="other_config" key="pmd-auto-lb-load-threshold"/>).
</p>
<p>
It uses current scheme of cycle based assignment of RX queues that
@ -690,6 +691,32 @@
once in few hours or a day or a week.
</p>
</column>
<column name="other_config" key="pmd-auto-lb-load-threshold"
type='{"type": "integer", "minInteger": 0, "maxInteger": 100}'>
<p>
Specifies the minimum PMD thread load threshold (% of used cycles) of
any non-isolated PMD threads when a PMD Auto Load Balance may be
triggered.
</p>
<p>
The default value is <code>95%</code>.
</p>
</column>
<column name="other_config" key="pmd-auto-lb-improvement-threshold"
type='{"type": "integer", "minInteger": 0, "maxInteger": 100}'>
<p>
Specifies the minimum evaluated % improvement in load distribution
across the non-isolated PMD threads that will allow a PMD Auto Load
Balance to occur.
</p>
<p>
Note, setting this parameter to 0 will always allow an auto load
balance to occur regardless of estimated improvement or not.
</p>
<p>
The default value is <code>25%</code>.
</p>
</column>
<column name="other_config" key="userspace-tso-enable"
type='{"type": "boolean"}'>
<p>