summaryrefslogtreecommitdiffstats
path: root/src/osd/OSD.cc
diff options
context:
space:
mode:
Diffstat (limited to 'src/osd/OSD.cc')
-rw-r--r--src/osd/OSD.cc148
1 files changed, 133 insertions, 15 deletions
diff --git a/src/osd/OSD.cc b/src/osd/OSD.cc
index ce46bb245ea..9c9e540cf61 100644
--- a/src/osd/OSD.cc
+++ b/src/osd/OSD.cc
@@ -37,6 +37,7 @@
#include "osd/PG.h"
#include "osd/scrubber/scrub_machine.h"
#include "osd/scrubber/pg_scrubber.h"
+#include "osd/ECCommon.h"
#include "include/types.h"
#include "include/compat.h"
@@ -504,6 +505,8 @@ void OSDService::shutdown_reserver()
void OSDService::shutdown()
{
+ pg_timer.stop();
+
mono_timer.suspend();
{
@@ -1064,7 +1067,12 @@ void OSDService::inc_osd_stat_repaired()
{
std::lock_guard l(stat_lock);
osd_stat.num_shards_repaired++;
- return;
+}
+
+void OSDService::set_osd_stat_repaired(int64_t count)
+{
+ std::lock_guard l(stat_lock);
+ osd_stat.num_shards_repaired = count;
}
float OSDService::compute_adjusted_ratio(osd_stat_t new_stat, float *pratio,
@@ -3219,6 +3227,11 @@ will start to track new ops received afterwards.";
scrub_purged_snaps();
}
+ else if (prefix == "clear_shards_repaired") {
+ int64_t count = cmd_getval_or<int64_t>(cmdmap, "count", 0);
+ service.set_osd_stat_repaired(count);
+ }
+
else if (prefix == "reset_purged_snaps_last") {
lock_guard l(osd_lock);
superblock.purged_snaps_last = 0;
@@ -4336,6 +4349,46 @@ void OSD::final_init()
"inject metadata error to an object");
ceph_assert(r == 0);
r = admin_socket->register_command(
+ "injectecreaderr " \
+ "name=pool,type=CephString " \
+ "name=objname,type=CephObjectname " \
+ "name=shardid,type=CephInt,req=true,range=0|255 " \
+ "name=type,type=CephInt,req=false " \
+ "name=when,type=CephInt,req=false " \
+ "name=duration,type=CephInt,req=false",
+ test_ops_hook,
+ "inject error for read of object in an EC pool");
+ ceph_assert(r == 0);
+ r = admin_socket->register_command(
+ "injectecclearreaderr " \
+ "name=pool,type=CephString " \
+ "name=objname,type=CephObjectname " \
+ "name=shardid,type=CephInt,req=true,range=0|255 " \
+ "name=type,type=CephInt,req=false",
+ test_ops_hook,
+ "clear read error injects for object in an EC pool");
+ ceph_assert(r == 0);
+ r = admin_socket->register_command(
+ "injectecwriteerr " \
+ "name=pool,type=CephString " \
+ "name=objname,type=CephObjectname " \
+ "name=shardid,type=CephInt,req=true,range=0|255 " \
+ "name=type,type=CephInt,req=false " \
+ "name=when,type=CephInt,req=false " \
+ "name=duration,type=CephInt,req=false",
+ test_ops_hook,
+ "inject error for write of object in an EC pool");
+ ceph_assert(r == 0);
+ r = admin_socket->register_command(
+ "injectecclearwriteerr " \
+ "name=pool,type=CephString " \
+ "name=objname,type=CephObjectname " \
+ "name=shardid,type=CephInt,req=true,range=0|255 " \
+ "name=type,type=CephInt,req=false",
+ test_ops_hook,
+ "clear write error inject for object in an EC pool");
+ ceph_assert(r == 0);
+ r = admin_socket->register_command(
"set_recovery_delay " \
"name=utime,type=CephInt,req=false",
test_ops_hook,
@@ -4440,6 +4493,12 @@ void OSD::final_init()
asok_hook,
"debug the scrubber");
ceph_assert(r == 0);
+ r = admin_socket->register_command(
+ "clear_shards_repaired "
+ "name=count,type=CephInt,req=false,range=0",
+ asok_hook,
+ "clear num_shards_repaired to clear health warning");
+ ceph_assert(r == 0);
// -- pg commands --
// old form: ceph pg <pgid> command ...
@@ -6469,8 +6528,10 @@ void TestOpsSocketHook::test_ops(OSDService *service, ObjectStore *store,
//directly request the osd make a change.
if (command == "setomapval" || command == "rmomapkey" ||
command == "setomapheader" || command == "getomap" ||
- command == "truncobj" || command == "injectmdataerr" ||
- command == "injectdataerr"
+ command == "truncobj" ||
+ command == "injectmdataerr" || command == "injectdataerr" ||
+ command == "injectecreaderr" || command == "injectecclearreaderr" ||
+ command == "injectecwriteerr" || command == "injectecclearwriteerr"
) {
pg_t rawpg;
int64_t pool;
@@ -6509,8 +6570,21 @@ void TestOpsSocketHook::test_ops(OSDService *service, ObjectStore *store,
ghobject_t gobj(obj, ghobject_t::NO_GEN, shard_id_t(uint8_t(shardid)));
spg_t pgid(curmap->raw_pg_to_pg(rawpg), shard_id_t(shardid));
if (curmap->pg_is_ec(rawpg)) {
- if ((command != "injectdataerr") && (command != "injectmdataerr")) {
- ss << "Must not call on ec pool, except injectdataerr or injectmdataerr";
+ if ((command != "injectdataerr") &&
+ (command != "injectmdataerr") &&
+ (command != "injectecreaderr") &&
+ (command != "injectecclearreaderr") &&
+ (command != "injectecwriteerr") &&
+ (command != "injectecclearwriteerr")) {
+ ss << "Must not call on ec pool";
+ return;
+ }
+ } else {
+ if ((command == "injectecreaderr") ||
+ (command == "injecteclearreaderr") ||
+ (command == "injectecwriteerr") ||
+ (command == "injecteclearwriteerr")) {
+ ss << "Only supported on ec pool";
return;
}
}
@@ -6589,6 +6663,38 @@ void TestOpsSocketHook::test_ops(OSDService *service, ObjectStore *store,
} else if (command == "injectmdataerr") {
store->inject_mdata_error(gobj);
ss << "ok";
+ } else if (command == "injectecreaderr") {
+ if (service->cct->_conf->bluestore_debug_inject_read_err) {
+ int64_t type = cmd_getval_or<int64_t>(cmdmap, "type", 0);
+ int64_t when = cmd_getval_or<int64_t>(cmdmap, "when", 0);
+ int64_t duration = cmd_getval_or<int64_t>(cmdmap, "duration", 1);
+ ss << ec_inject_read_error(gobj, type, when, duration);
+ } else {
+ ss << "bluestore_debug_inject_read_err not enabled";
+ }
+ } else if (command == "injectecclearreaderr") {
+ if (service->cct->_conf->bluestore_debug_inject_read_err) {
+ int64_t type = cmd_getval_or<int64_t>(cmdmap, "type", 0);
+ ss << ec_inject_clear_read_error(gobj, type);
+ } else {
+ ss << "bluestore_debug_inject_read_err not enabled";
+ }
+ } else if (command == "injectecwriteerr") {
+ if (service->cct->_conf->bluestore_debug_inject_read_err) {
+ int64_t type = cmd_getval_or<int64_t>(cmdmap, "type", 0);
+ int64_t when = cmd_getval_or<int64_t>(cmdmap, "when", 0);
+ int64_t duration = cmd_getval_or<int64_t>(cmdmap, "duration", 1);
+ ss << ec_inject_write_error(gobj, type, when, duration);
+ } else {
+ ss << "bluestore_debug_inject_read_err not enabled";
+ }
+ } else if (command == "injectecclearwriteerr") {
+ if (service->cct->_conf->bluestore_debug_inject_read_err) {
+ int64_t type = cmd_getval_or<int64_t>(cmdmap, "type", 0);
+ ss << ec_inject_clear_write_error(gobj, type);
+ } else {
+ ss << "bluestore_debug_inject_read_err not enabled";
+ }
}
return;
}
@@ -9485,7 +9591,8 @@ void OSD::handle_pg_query_nopg(const MQuery& q)
q.query.epoch_sent,
osdmap->get_epoch(),
empty,
- PastIntervals()};
+ PastIntervals(),
+ PG_FEATURE_CLASSIC_ALL};
m = new MOSDPGNotify2(spg_t{pgid.pgid, q.query.from},
std::move(notify));
}
@@ -9939,7 +10046,8 @@ const char** OSD::get_tracked_conf_keys() const
"osd_scrub_max_interval",
"osd_op_thread_timeout",
"osd_op_thread_suicide_timeout",
- NULL
+ "osd_max_scrubs",
+ nullptr
};
return KEYS;
}
@@ -9983,6 +10091,10 @@ void OSD::handle_conf_change(const ConfigProxy& conf,
service.snap_reserver.set_max(cct->_conf->osd_max_trimming_pgs);
}
if (changed.count("osd_max_scrubs")) {
+ dout(0) << fmt::format(
+ "{}: scrub concurrency max changed to {}",
+ __func__, cct->_conf->osd_max_scrubs)
+ << dendl;
service.scrub_reserver.set_max(cct->_conf->osd_max_scrubs);
}
if (changed.count("osd_op_complaint_time") ||
@@ -10154,22 +10266,28 @@ void OSD::maybe_override_max_osd_capacity_for_qos()
<< dendl;
// Get the threshold IOPS set for the underlying hdd/ssd.
- double threshold_iops = 0.0;
+ double hi_threshold_iops = 0.0;
+ double lo_threshold_iops = 0.0;
if (store_is_rotational) {
- threshold_iops = cct->_conf.get_val<double>(
+ hi_threshold_iops = cct->_conf.get_val<double>(
"osd_mclock_iops_capacity_threshold_hdd");
+ lo_threshold_iops = cct->_conf.get_val<double>(
+ "osd_mclock_iops_capacity_low_threshold_hdd");
} else {
- threshold_iops = cct->_conf.get_val<double>(
+ hi_threshold_iops = cct->_conf.get_val<double>(
"osd_mclock_iops_capacity_threshold_ssd");
+ lo_threshold_iops = cct->_conf.get_val<double>(
+ "osd_mclock_iops_capacity_low_threshold_ssd");
}
// Persist the iops value to the MON store or throw cluster warning
- // if the measured iops exceeds the set threshold. If the iops exceed
- // the threshold, the default value is used.
- if (iops > threshold_iops) {
+ // if the measured iops is not in the threshold range. If the iops is
+ // not within the threshold range, the current/default value is retained.
+ if (iops < lo_threshold_iops || iops > hi_threshold_iops) {
clog->warn() << "OSD bench result of " << std::to_string(iops)
- << " IOPS exceeded the threshold limit of "
- << std::to_string(threshold_iops) << " IOPS for osd."
+ << " IOPS is not within the threshold limit range of "
+ << std::to_string(lo_threshold_iops) << " IOPS and "
+ << std::to_string(hi_threshold_iops) << " IOPS for osd."
<< std::to_string(whoami) << ". IOPS capacity is unchanged"
<< " at " << std::to_string(cur_iops) << " IOPS. The"
<< " recommendation is to establish the osd's IOPS capacity"
generated by cgit v1.2.3 (git 2.39.1) at 2025-01-18 11:22:49 +0100