# 8.0.27, and 8.0.29 unable to restore backup taken from 8.0.26

**URL:** <https://forums.percona.com/t/8-0-27-and-8-0-29-unable-to-restore-backup-taken-from-8-0-26/18723>\
**Category:** Percona Server for MySQL 8.0\
**Created:** [November 23, 2022, 12:22am UTC](https://forums.percona.com/t/8-0-27-and-8-0-29-unable-to-restore-backup-taken-from-8-0-26/18723 "2022-11-23T00:22:37Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![Vamsi\_P](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/vamsi_p/32/2549_2.png) [@Vamsi\_P](https://forums.percona.com/u/Vamsi_P)\
**Post date:** [November 23, 2022, 12:22am UTC](https://forums.percona.com/t/8-0-27-and-8-0-29-unable-to-restore-backup-taken-from-8-0-26/18723/1 "2022-11-23T00:22:38Z")

</div>

hi Folks,  
We were using  
percona-server-common=8.0.26-17-2.bionic  
percona-server-server=8.0.26-17-2.bionic  
percona-server-client=8.0.26-17-2.bionic  
percona-xtrabackup-80=8.0.26-18-1.bionic

That build seems to have been removed as we are now getting  
`E: Version '8.0.26-18-1.bionic' for 'percona-xtrabackup-80' was not found` when we try to install it on new host.

So we tried using 8.0.27 MyRocks and Xtrabackup on a new host to restore from backup taken by earlier build (8.0.26), and it seems to crash when it tries to “start slave”. This is the stack trace we found in error log. We see same stack trace even if we use 8.0.29 MyRocks and Xtrabackup as well.

```auto
11:00:17 UTC - mysqld got signal 11 ;
Most likely, you have hit a bug, but this error can also be caused by malfunctioning hardware.

Build ID: a9fa9e3f0b519ac6b75e0806a6e6e156bb57c4be
Server Version: 8.0.27-18 Percona Server (GPL), Release '18', Revision '24801e21b45'

Thread pointer: 0x7f68d1419000
Attempting backtrace. You can use the following information to find out
where mysqld died. If you see no messages after this, something went
terribly wrong...

/lib/x86_64-linux-gnu/libc.so.6(clone+0x3f) [0x7f68f864d61f]
stack_bottom = 7f68e8d4dd58 thread_stack 0x40000
/usr/sbin/mysqld(my_print_stacktrace(unsigned char const*, unsigned long)+0x3d) [0x55f72cf76f5d]*
*/usr/sbin/mysqld(handle_fatal_signal+0x3c3) [0x55f72c1d7e83]*
*/lib/x86_64-linux-gnu/libpthread.so.0(+0x12980) [0x7f68fa4e9980]*
*/usr/lib/mysql/plugin/ha_rocksdb.so(myrocks::rdb_should_hide_ttl_rec(myrocks::Rdb_key_def const&, rocksdb::Slice const&, myrocks::Rdb_transaction*)+0x79) [0x7f68d5b5a657]
/usr/lib/mysql/plugin/ha_rocksdb.so(myrocks::Rdb_iterator_base::get(rocksdb::Slice const*, rocksdb::PinnableSlice*, myrocks::Rdb_lock_type, bool, bool)+0x1bb) [0x7f68d5b960cb]
/usr/lib/mysql/plugin/ha_rocksdb.so(myrocks::ha_rocksdb::check_and_lock_unique_pk(myrocks::ha_rocksdb::update_row_info const&, bool*, bool)+0xc5) [0x7f68d5b5b8a9]*
*/usr/lib/mysql/plugin/ha_rocksdb.so(myrocks::ha_rocksdb::check_uniqueness_and_lock(myrocks::ha_rocksdb::update_row_info const&, bool, bool)+0x80) [0x7f68d5b7fc0e]*
*/usr/lib/mysql/plugin/ha_rocksdb.so(myrocks::ha_rocksdb::update_write_row(unsigned char const*, unsigned char const*, bool)+0x1f7) [0x7f68d5b804d5]*
*/usr/lib/mysql/plugin/ha_rocksdb.so(myrocks::ha_rocksdb::write_row(unsigned char*)+0x5d) [0x7f68d5b80545]
/usr/sbin/mysqld(handler::ha_write_row(unsigned char*)+0x1a4) [0x55f72bd93b44]*
*/usr/sbin/mysqld(Write_rows_log_event::write_row(Relay_log_info const*, bool)+0xfd) [0x55f72cc137cd]
/usr/sbin/mysqld(Write_rows_log_event::do_exec_row(Relay_log_info const*)+0x19) [0x55f72cc13b89]*
*/usr/sbin/mysqld(Rows_log_event::do_apply_row(Relay_log_info const*)+0x26) [0x55f72cc00076]
/usr/sbin/mysqld(Rows_log_event::do_apply_event(Relay_log_info const*)+0x1353) [0x55f72cc1f7f3]*
*/usr/sbin/mysqld(slave_worker_exec_job_group(Slave_worker*, Relay_log_info*)+0x171) [0x55f72cc9e1f1]
/usr/sbin/mysqld(+0x1e9b50b) [0x55f72cca550b]
/usr/sbin/mysqld(+0x2680e34) [0x55f72d48ae34]
/lib/x86_64-linux-gnu/libpthread.so.0(+0x76db) [0x7f68fa4de6db]

```

Any clue on why 8.0.27 and 8.0.29 are unable to restore backup taken from older build 8.0.26?

Thanks,  
Vamsi

---

<div class="post-metadata">

**Author:** ![satya.bodapati](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/satya.bodapati/32/1322_2.png) [@satya.bodapati](https://forums.percona.com/u/satya.bodapati)\
**Post date:** [November 23, 2022, 10:10am UTC](https://forums.percona.com/t/8-0-27-and-8-0-29-unable-to-restore-backup-taken-from-8-0-26/18723/2 "2022-11-23T10:10:43Z")

</div>

> [@Vamsi\_P](#):
>
> `rdb_should_hide_ttl_rec`

Can you please check if you are facing this rocksdb issue?

> <https://github.com/facebook/mysql-5.6/pull/1192>
>
> \*\*Problem:\*\*
> 
> This execution:
> 
> \`\`\`
> CREATE TABLE t1 (
> \`id\` int NOT NULL AUT…O\_INCREMENT,
> \`username\` varchar(64) NOT NULL,
> \`date\` date NOT NULL,
> \`hour\` tinyint unsigned NOT NULL,
> \`ipv4\` varchar(15) NOT NULL,
> \`ipv6\` varchar(39) NOT NULL,
> \`acctinputoctets\` bigint DEFAULT NULL,
> \`acctoutputoctets\` bigint DEFAULT NULL,
> PRIMARY KEY (\`id\`),
> UNIQUE KEY \`username\_day\_hour\` (\`username\`,\`date\`,\`hour\`)
> ) ENGINE=ROCKSDB AUTO\_INCREMENT=1 DEFAULT CHARSET=latin1 COMMENT='ttl\_duration=604800;';
> 
> INSERT INTO t1 (\`username\`, \`date\`, \`hour\`, \`ipv4\`, \`ipv6\`, \`acctinputoctets\`, \`acctoutputoctets\`)
> VALUES ('100002992184', '2022-06-08', '10', '192.168.1.70', '', '1', '2');
> \`\`\`
> 
> causes the following crash:
> 
> \`\`\`
> 13:12:26 UTC - mysqld got signal 11 ;
> Most likely, you have hit a bug, but this error can also be caused by malfunctioning hardware.
> Thread pointer: 0x7fb21ec0b7c0
> Attempting backtrace. You can use the following information to find out
> where mysqld died. If you see no messages after this, something went
> terribly wrong...
> stack\_bottom = 7fb24fb67f08 thread\_stack 0x46000
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(my\_print\_stacktrace(unsigned char const\*, unsigned long)+0x42) \[0x726eaa2\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(handle\_fatal\_signal+0x232) \[0x5fa7132\]
> /lib/x86\_64-linux-gnu/libpthread.so.0(+0x14420) \[0x7fb25ead0420\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(rocksdb::Slice::data() const+0xc) \[0x7c4fa7c\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::Rdb\_string\_reader::Rdb\_string\_reader(rocksdb::Slice const\*)+0x21) \[0x7c4f4f1\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::rdb\_should\_hide\_ttl\_rec(myrocks::Rdb\_key\_def const&, rocksdb::Slice const&, myrocks::Rdb\_transaction\*)+0x101) \[0x7c0d6f1\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::Rdb\_iterator\_base::get(rocksdb::Slice const\*, rocksdb::PinnableSlice\*, myrocks::Rdb\_lock\_type, bool, bool)+0x32b) \[0x83e4ffb\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::ha\_rocksdb::check\_and\_lock\_sk(unsigned int, myrocks::ha\_rocksdb::update\_row\_info const&, bool\*, bool)+0x61a) \[0x7c1ac7a\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::ha\_rocksdb::check\_uniqueness\_and\_lock(myrocks::ha\_rocksdb::update\_row\_info const&, bool, bool)+0x190) \[0x7c1b140\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::ha\_rocksdb::update\_write\_row(unsigned char const\*, unsigned char const\*, bool)+0x271) \[0x7c19941\]
> /home/ldonoso/src/fb-mysql-bld/runtime\_output\_directory/mysqld(myrocks::ha\_rocksdb::write\_row(unsigned char\*)+0x2e5) \[0x7c195d5\]
> \`\`\`
> 
> \*\*Solution\*\*:
> 
> If \`all\_parts\_used\` is \`false\` we can skip the check of the ttl of the row.
> 
> It is not a problem if the record is stale since, next, \`iter.seek\` is called and \`iter.seek\` checks the TTL.

May be it is not related to backup

---

<div class="post-metadata">

**Author:** ![Vamsi\_P](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/vamsi_p/32/2549_2.png) [@Vamsi\_P](https://forums.percona.com/u/Vamsi_P)\
**Post date:** [November 29, 2022, 12:31pm UTC](https://forums.percona.com/t/8-0-27-and-8-0-29-unable-to-restore-backup-taken-from-8-0-26/18723/3 "2022-11-29T12:31:25Z")

</div>

Thanks a lot Satya! Yes it is the issue. It seems like that issue has been fixed in FB branch? Any estimate of when it will make it to Percona MyRocks?

---

<div class="post-metadata">

**Author:** ![Vamsi\_P](https://sea1.discourse-cdn.com/flex019/user_avatar/forums.percona.com/vamsi_p/32/2549_2.png) [@Vamsi\_P](https://forums.percona.com/u/Vamsi_P)\
**Post date:** [February 14, 2023, 6:20pm UTC](https://forums.percona.com/t/8-0-27-and-8-0-29-unable-to-restore-backup-taken-from-8-0-26/18723/4 "2023-02-14T18:20:24Z")

</div>

@satya.bodapati has the fix been ported to Percona MyRocks yet? if so, which build? Thanks!
