-
Type: Bug
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: None
When running test_tiered_abort -T 12 -t 10 sometimes the child process runs until killed by the parent and when the parent opens the database to verify, running recovery from wiredtiger_open will fail with:
[1631713589:161622][23222:0x7ff0b0a407c0], tiered:local, txn-recover: void __wt_dhandle_update_write_gens(WT_SESSION_IMPL *), 994: btree != NULL [1631713589:161640][23222:0x7ff0b0a407c0], tiered:local, txn-recover: void __wt_abort(WT_SESSION_IMPL *), 28: aborting WiredTiger library
Here is the stack. The reason is that the dhandle is the tiered dhandle:
#2 0x00000000006eb9c4 in __wt_abort (session=0x6300000013c0) at ../src/os_common/os_abort.c:30 #3 0x0000000000558d39 in __wt_dhandle_update_write_gens (session=0x6300000013c0) at ../src/conn/conn_dhandle.c:994 #4 0x00000000008e78b2 in __wt_txn_recover (session=0x6300000013c0, cfg=0x7ffd8a34f9a0) at ../src/txn/txn_recover.c:1031 #5 0x000000000056ebe5 in __wt_connection_workers (session=0x630000000400, cfg=0x7ffd8a34f9a0) at ../src/conn/conn_open.c:220 #6 0x0000000000514057 in wiredtiger_open (home=0x0, event_handler=0x0, config=0x7ffd8a351950 "log=(archive=false,recover=on),extensions=(/mnt/data0/sue/wt.7957/build_posix/ext/storage_sources/local_store/.libs/libwiredtiger_local_store.so=(early_load=true))", connectionp=0x7ffd8a351060) at ../src/conn/conn_api.c:2904 #7 0x00000000004f7f11 in main (argc=0, argv=0x7ffd8a3527a8) at ../../../test/csuite/tiered_abort/main.c:723 (gdb) p dhandle->name $1 = 0x6020004ba3d0 "tier:shadow" (gdb) p dhandle->handle $2 = (void *) 0x0