View the thread stack according to the lwtid of opengauss/mogdb.

stay openGauss/MogDB database THREAD_WAIT_STATUS、PG_THREAD_WAIT_STATUS、GLOBAL_THREAD_WAIT_STATUS、LOCAL_ACTIVE_SESSION There are several views in lwtid This column , It is the lightweight thread number of the current thread . You can pass through according to this column pstack Command to grab the stack , Analysis of supply problems .

As shown below , It can be used pg_stat_activity and dbe_perf.thread_wait_status combination , Query some currently running SQL, adopt

MogDB=# select  pid,lwtid,state,wait_event,query from pg_stat_activity a,dbe_perf.thread_wait_status s where a.pid=s.tid;      pid       |  lwtid  | state  | wait_event |                                                       query                                                       ----------------+---------+--------+------------+------------------------------------------------------------------------------------------------------------------- 23043933468416 | 1724740 | active | none       | select  pid,lwtid,state,wait_event,query from pg_stat_activity a,dbe_perf.thread_wait_status s where a.pid=s.tid; 23044103272192 | 1724300 | active | none       | select pg_sleep(30); 23044453037824 | 1526848 |        | none       |  23044479776512 | 1526847 |        | none       |  23044508612352 | 1526846 | active | none       | WLM fetch collect info from data nodes 23044674287360 | 1526842 | idle   | none       |  23044771804928 | 1526839 | idle   | none       |  23044739823360 | 1526840 | active | none       |  23044698928896 | 1526841 | active | none       |  23044832098048 | 1526837 |        | none       |  23044928567040 | 1526834 | active | none       | (11 rows)

We go through pstack Go check the top lwtid=1724300 Corresponding select pg_sleep(30); The stack .

[[email protected] ~]$ pstack 1724300Thread 1 (Thread 0x14f55f96f700 (LWP 1724300)):#0  0x000014f98e8d829f in select () from /lib64/libc.so.6#1  0x0000555ba97ae25d in pg_usleep(long) ()#2  0x0000555ba87cef4a in pg_sleep(FunctionCallInfoData*) ()#3  0x0000555ba8fad525 in ?? ()#4  0x0000555ba8fadb59 in ExecProject(ProjectionInfo*, ExprDoneCond*) ()#5  0x0000555ba8ffb948 in ExecResult(ResultState*) ()#6  0x0000555ba8f9c6ee in ExecProcNode(PlanState*) ()#7  0x0000555ba8f95318 in standard_ExecutorRun(QueryDesc*, ScanDirection, long) ()#8  0x0000555ba8e25ed5 in ?? ()#9  0x0000555ba8f956ad in ExecutorRun(QueryDesc*, ScanDirection, long) ()#10 0x0000555ba8f148b3 in ?? ()#11 0x0000555ba8f14fa0 in PortalRun(PortalData*, long, bool, _DestReceiver*, _DestReceiver*, char*) ()#12 0x0000555ba8f095a2 in ?? ()#13 0x0000555ba8f0ce97 in PostgresMain(int, char**, char const*, char const*) ()#14 0x0000555ba8e6b501 in ?? ()#15 0x0000555ba8e89fb8 in int GaussDbThreadMain<(knl_thread_role)1>(knl_thread_arg*) ()#16 0x0000555ba8e6b585 in ?? ()#17 0x000014f98ebb117a in start_thread () from /lib64/libpthread.so.0#18 0x000014f98e8e0dc3 in clone () from /lib64/libc.so.6

Can be seen clearly , Stack from PostgresMain Later, I called pg_sleep Every part of , And later pg_usleep Dormant part .

It can also be based on lwtid Find some key threads you want to analyze , For example, some threads with special functions , Such as WDRSnapshot、ASP、CheckPointer etc. 、 Or it is running for a long time in the database SQL And the thread corresponding to the waiting event that needs to be analyzed .

 node_name | db_name  |      thread_name       |     query_id     |      tid       |   sessionid    |  lwtid  | psessionid | tlevel | smpid | wait_status | wait_event | locktag | lockmode | block_sessionid | global_sessionid -----------+----------+------------------------+------------------+----------------+----------------+---------+------------+--------+-------+-------------+------------+---------+----------+-----------------+------------------ dn_6001   | postgres | gsql                   | 1407374883636205 | 23044103272192 | 23044103272192 | 1724935 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | WLMArbiter             |                0 | 23044453037824 | 23044453037824 | 1526848 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | WorkloadMonitor        |                0 | 23044479776512 | 23044479776512 | 1526847 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | workload               |                0 | 23044508612352 | 23044508612352 | 1526846 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | statement flush thread |                0 | 23044674287360 | 23044674287360 | 1526842 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | WDRSnapshot            | 1407374883553280 | 23044771804928 | 23044771804928 | 1526839 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | PercentileJob          |                0 | 23044739823360 | 23044739823360 | 1526840 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | Asp                    |                0 | 23044698928896 | 23044698928896 | 1526841 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | ApplyLauncher          |                0 | 23044832098048 | 23044832098048 | 1526837 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0 dn_6001   | postgres | JobScheduler           |                0 | 23044928567040 | 23044928567040 | 1526834 |            |      0 |     0 | none        | none       |         |          |                 | 0:0#0(10 rows)

As shown below , You can see lwtid=1526839 The corresponding is WDRSnapshot The current stack condition of the thread .

[[email protected] ~]$ pstack 1526839Thread 1 (Thread 0x14f5876ff700 (LWP 1526839)):#0  0x000014f98e8d829f in select () from /lib64/libc.so.6#1  0x0000555ba97ae25d in pg_usleep(long) ()#2  0x0000555ba8a83e83 in SnapshotNameSpace::SubSnapshotMain() ()#3  0x0000555ba8a843c5 in SnapshotMain() ()#4  0x0000555ba8e87e41 in int GaussDbThreadMain<(knl_thread_role)35>(knl_thread_arg*) ()#5  0x0000555ba8e6b585 in ?? ()#6  0x000014f98ebb117a in start_thread () from /lib64/libpthread.so.0#7  0x000014f98e8e0dc3 in clone () from /lib64/libc.so.6