Spent a little time investigating this last week with a few small cases and haven't reproduced it yet.

This is likely to be solved by the ThreadPool leak fixes.

I will open this one b/c WriteEngine leaks even with jemalloc and ThreadPool fixes.
The problem caused by CSC::fCatalogMap that nobody cleans up. The WriteEngine memory consumption raises without limits. There could be other victims of this fCatalogMap.
There are two ways: either use vacuum background cleaning or clean the session on disconnect.
This map could affects the overal performance when WE is up long enough.

Please review the fix.

For QA. You could monitor the memory leak with top. It is easy to see the difference b/w memory consumption in the beggining and after 300 mixed DML operations: insert, update, delete.

Build verified: 1.2.4-1 nightly

Executed Autopilot's concurrentDML test on both 1.2.3-1 and 1.2.4-1. Write engine on 1.2.4-1 used 1/4 of the memory 1.2.3-1 used, 4.5% instead 18.2%.

server commit:
137b9a8
engine commit:
5882808

1.2.3-1

Before test

top - 17:39:37 up 1 min, 1 user, load average: 0.39, 0.20, 0.07
Tasks: 131 total, 1 running, 130 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.5 us, 0.3 sy, 0.0 ni, 99.1 id, 0.0 wa, 0.0 hi, 0.1 si, 0.0 st
KiB Mem : 2914036 total, 1925024 free, 571840 used, 417172 buff/cache
KiB Swap: 1572860 total, 1572860 free, 0 used. 2142440 avail Mem

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
13395 mysql 20 0 2493404 172924 18788 S 0.0 5.9 0:00.31 mysqld
13627 root 19 -1 2158680 81636 10188 S 0.0 2.8 0:00.09 PrimProc
13739 root 19 -1 332448 69668 8608 S 0.0 2.4 0:00.08 ExeMgr
3305 root 20 0 713676 45216 12088 S 0.7 1.6 0:39.49 ProcMon
13515 root 20 0 246384 40696 8312 S 0.3 1.4 0:00.08 controllernode
13559 root 20 0 241352 40268 16056 S 0.0 1.4 0:00.07 workernode
13934 root 20 0 266448 33096 15148 S 0.0 1.1 0:00.14 DMLProc
13088 root 20 0 362680 32988 9060 S 0.7 1.1 0:01.09 ProcMgr
13831 root 20 0 429148 30368 10368 S 0.0 1.0 0:00.10 WriteEngineServ
13866 root 20 0 224388 28200 9980 S 0.0 1.0 0:00.09 DDLProc
13535 root 20 0 197308 24360 8352 S 0.0 0.8 0:00.09 ServerMonitor

After test

top - 20:54:25 up 3:16, 1 user, load average: 1.54, 2.54, 2.42
Tasks: 131 total, 1 running, 130 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.3 us, 0.3 sy, 0.0 ni, 99.2 id, 0.1 wa, 0.0 hi, 0.1 si, 0.0 st
KiB Mem : 2914036 total, 960784 free, 1709152 used, 244100 buff/cache
KiB Swap: 1572860 total, 1333748 free, 239112 used. 969824 avail Mem

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
13831 root 20 0 2786176 529816 12920 S 0.0 18.2 70:15.11 WriteEngineServ
13627 root 19 -1 2395716 426884 10196 S 0.0 14.6 2:10.43 PrimProc
13559 root 20 0 596128 191864 12956 S 0.0 6.6 4:08.46 workernode
13395 mysql 20 0 3029552 162148 6340 S 0.0 5.6 19:08.07 mysqld
13739 root 19 -1 2319392 89148 4568 S 0.0 3.1 0:49.38 ExeMgr
13515 root 20 0 537324 60280 2984 S 0.0 2.1 4:47.66 controllernode
13866 root 20 0 576752 48028 3960 S 0.0 1.6 2:00.86 DDLProc
13934 root 20 0 1648552 41896 9324 S 0.3 1.4 8:15.09 DMLProc

1.2.4-1

top - 17:36:52 up 1 min, 1 user, load average: 0.31, 0.17, 0.06
Tasks: 130 total, 1 running, 129 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.0 us, 0.1 sy, 0.0 ni, 99.9 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem : 2914036 total, 1977348 free, 570332 used, 366356 buff/cache
KiB Swap: 1572860 total, 1572860 free, 0 used. 2147896 avail Mem

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4066 mysql 20 0 2497544 168316 18184 S 0.0 5.8 0:00.25 mysqld
4299 root 19 -1 2162820 81592 10112 S 0.0 2.8 0:00.09 PrimProc
4389 root 19 -1 336592 69700 8620 S 0.0 2.4 0:00.09 ExeMgr
3193 root 20 0 717820 45296 12128 S 0.0 1.6 0:40.87 ProcMon
4186 root 20 0 262820 44868 8376 S 0.0 1.5 0:00.07 controllernode
4265 root 20 0 245496 38272 16084 S 0.0 1.3 0:00.07 workernode
3759 root 20 0 366824 33168 9100 S 0.0 1.1 0:01.19 ProcMgr
4592 root 20 0 270588 33156 15152 S 0.0 1.1 0:00.15 DMLProc
4206 root 20 0 226036 30772 8644 S 0.0 1.1 0:00.18 ServerMonitor
4478 root 20 0 433288 30268 10148 S 0.0 1.0 0:00.09 WriteEngineServ
4548 root 20 0 228528 28240 10008 S 0.0 1.0 0:00.08 DDLProc
3114 root 20 0 573920 17244 6108 S 0.0 0.6 0:00.24 tuned

After test

top - 12:57:04 up 19:21, 1 user, load average: 0.16, 0.05, 0.06
Tasks: 126 total, 1 running, 125 sleeping, 0 stopped, 0 zombie
%Cpu(s): 0.0 us, 0.1 sy, 0.0 ni, 99.9 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st
KiB Mem : 2914036 total, 1288248 free, 1345764 used, 280024 buff/cache
KiB Swap: 1572860 total, 1359868 free, 212992 used. 1327712 avail Mem

PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4299 root 19 -1 2408048 427424 10500 S 0.3 14.7 3:10.69 PrimProc
4265 root 20 0 600272 187736 13328 S 0.0 6.4 4:21.87 workernode
4066 mysql 20 0 3033692 171432 7116 S 0.3 5.9 21:54.91 mysqld
4478 root 20 0 2605992 129896 14244 S 0.0 4.5 69:05.14 WriteEngineServ
4389 root 19 -1 2356320 93852 5060 S 0.0 3.2 1:15.51 ExeMgr
4186 root 20 0 557860 59552 3300 S 0.0 2.0 5:26.83 controllernode
4592 root 20 0 2050056 54056 10592 S 0.0 1.9 8:55.58 DMLProc
4548 root 20 0 650524 44144 5220 S 0.0 1.5 2:09.68 DDLProc
3193 root 20 0 734212 23996 6556 S 0.0 0.8 6:09.86 ProcMon
3759 root 20 0 366824 22300 3544 S 0.0 0.8 0:29.11 ProcMgr

Will this fix be available on version 1.1.8?

Yes, we plan to add this fix into 1.1

Any forecast about when this will be released on 1.1? I don't see it on this page:
https://jira.mariadb.org/projects/MCOL/versions/23206

It's already backported in 1.1 branch so 1.1.8 contains the fix.