After restart indexer makes the block skipped more than expected

[lichao-cobo]

My ton indexer start index from 39647000

root@ton:~# docker top ton-indexer-index-worker-1
UID                 PID                 PPID                C                   STIME               TTY                 TIME                CMD
root                1688324             1688305             0                   03:30               ?                   00:00:00            /bin/bash /entrypoint.sh --from 39647000 --max-insert-actors 12 --max-active-tasks 2048
root                1688370             1688324             99                  03:30               ?                   00:12:06            ton-index-postgres --host 172.19.0.2 --port 5432 --user postgres --password changeme --dbname ton_index --db /tondb --from 39647000 --max-insert-actors 12 --max-active-tasks 2048

And after one of restart process the indexer docker log shows it starts index from 39977285 which more than 2000 blocks higher than 39975375 = 39647000 + 328375 (Skipped 328375 existing seqnos)

root@ton:~# docker logs -f --tail 20 ton-indexer-index-worker-1
Postgres password file: /run/secrets/postgres_password
Postgres host: postgres (ip: 172.19.0.2)
[ 3][t 1][2024-08-27 03:30:02.977536668][InsertManagerBase.cpp:18][!insertmanager]	InsertManagerBase::start_up called
[ 3][t 1][2024-08-27 03:30:02.977632360][InsertManagerBase.cpp:28][!insertmanager]	Insert manager(parallel=12, max_batch_mc_blocks=2000, max_batch_blocks=2000, max_batch_txs=10000, max_batch_msgs=10000)
[ 3][t 2][2024-08-27 03:30:02.978696631][InsertManagerPostgres.cpp:1148][!insertmanager]	Reading existing seqnos
[ 3][t 2][2024-08-27 03:30:03.397618896][IndexScheduler.cpp:76][!indexscheduler]	Found 328375 existing seqnos
[ 3][t 3][2024-08-27 03:30:04.423969716][IndexScheduler.cpp:93][!indexscheduler]	Skipped 328375 existing seqnos
[ 3][t 2][2024-08-27 03:30:11.877006037][InsertManagerBase.cpp:81][!insertmanager]	Inserting batch[mb=6, b=6, txs=18, msgs=6, traces=0]
[ 3][t 4][2024-08-27 03:30:13.876317754][InsertManagerBase.cpp:81][!insertmanager]	Inserting batch[mb=1, b=1, txs=3, msgs=1, traces=0]
[ 3][t 1][2024-08-27 03:30:14.761739278][IndexScheduler.cpp:174][!indexscheduler]	Last: 39977285 / 39983362	Blk/s: 0.640000	ETA: 2h 38m 15s	Q[1M, 1b, 3t, 1m]
[ 3][t 7][2024-08-27 03:30:22.110379743][InsertManagerBase.cpp:81][!insertmanager]	Inserting batch[mb=1, b=1, txs=3, msgs=1, traces=0]
[ 3][t 7][2024-08-27 03:30:36.162721056][IndexScheduler.cpp:174][!indexscheduler]	Last: 39977285 / 39983362	Blk/s: 0.466560	ETA: 3h 37m 5s	Q[3M, 3b, 9t, 3m]
[ 2][t 6][2024-08-27 03:30:46.512964476][InterfaceDetectors.hpp:939][!nft_item_detector]	Failed to parse content for 0:1CC2F1AE681FAB568FECCE631DA2B19EC70039BF7A9EE91CD6FC2808D2C4915E: [Error : 0 : Unknown token data type] Content: te6cckEBAQEAAgAAAEysuc0=

And the indexer api height shows the same result

[lichao-cobo]

think may caused by --max-active-tasks 2048
after remove it, back to

[ 3][t 5][2024-08-27 03:56:48.839786142][IndexScheduler.cpp:174][!indexscheduler]	Last: 39975383 / 39983518	Blk/s: 0.300000	ETA: 7h 31m 56s	Q[2M, 103b, 9655t, 15083m]