[enhancement](recover) support skipping missing version in select by session variable

[xy720]

Proposed changes

Recently, in version 1.2.7, we encountered a situation where all replicas of tablets in table are having missing versions. In this case, FE is unable to recover these tablets, and the table is unable to select.

In pr #18986, I noticed that we have support config recover_with_skip_missing_version to ignore the visible version in FE partition.

This config only work when the replica version is behind partition version. But if the replica on be has a version missing, it actually does not skip any missing versions on be.

For example,

1、if the replica versions in BE is {1-100, 101, 102, 103}, the partition visible version is 100, recover_with_skip_missing_version = ignore_version is working well.

2、if the replica versions in BE is {1-100, 103}, the partition visible version is 100, recover_with_skip_missing_version = ignore_version is not working.

This commit support session variable skip_missing_version to control the query behavior.

If skip_missing_version is set to true, the query will always try to select the one with the highest lastSuccessVersion among all surviving BE replicas.

If skip_missing_version is set to true, the query will always skip the missing rowsets in BE and only return the data from existing rowsets.

Further comments

If this is a relatively large or complex change, kick off the discussion at dev@doris.apache.org by explaining why you chose the solution you did and what alternatives you considered, etc...

[xy720]

run buildall

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[xy720]

run buildall

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[doris-robot]

(From new machine)TeamCity pipeline, clickbench performance test result:
the sum of best hot time: 46.54 seconds
stream load tsv: 556 seconds loaded 74807831229 Bytes, about 128 MB/s
stream load json: 20 seconds loaded 2358488459 Bytes, about 112 MB/s
stream load orc: 65 seconds loaded 1101869774 Bytes, about 16 MB/s
stream load parquet: 31 seconds loaded 861443392 Bytes, about 26 MB/s
insert into select: 29.0 seconds inserted 10000000 Rows, about 344K ops/s
storage size: 17162156215 Bytes

[doris-robot]

(From new machine)TeamCity pipeline, clickbench performance test result:
the sum of best hot time: 46.09 seconds
stream load tsv: 554 seconds loaded 74807831229 Bytes, about 128 MB/s
stream load json: 20 seconds loaded 2358488459 Bytes, about 112 MB/s
stream load orc: 65 seconds loaded 1101869774 Bytes, about 16 MB/s
stream load parquet: 32 seconds loaded 861443392 Bytes, about 25 MB/s
insert into select: 29.0 seconds inserted 10000000 Rows, about 344K ops/s
storage size: 17162115673 Bytes

[xy720]

run buildall

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[doris-robot]

TeamCity be ut coverage result:
Function Coverage: 36.97% (8283/22405)
Line Coverage: 29.06% (66436/228624)
Region Coverage: 27.72% (34499/124438)
Branch Coverage: 24.33% (17522/72020)
Coverage Report: http://coverage.selectdb-in.cc/coverage/6a64c9df2272c78a4f44104259ea6fac04d6276c_6a64c9df2272c78a4f44104259ea6fac04d6276c/report/index.html

[doris-robot]

(From new machine)TeamCity pipeline, clickbench performance test result:
the sum of best hot time: 46.77 seconds
stream load tsv: 556 seconds loaded 74807831229 Bytes, about 128 MB/s
stream load json: 20 seconds loaded 2358488459 Bytes, about 112 MB/s
stream load orc: 65 seconds loaded 1101869774 Bytes, about 16 MB/s
stream load parquet: 32 seconds loaded 861443392 Bytes, about 25 MB/s
insert into select: 29.2 seconds inserted 10000000 Rows, about 342K ops/s
storage size: 17161981198 Bytes

[xy720]

run p0

[dataroaring]

We'd better use a session variable to control behavior of query. Recovery is different than query, recovery is a action without lost, however, if query ignores some missing versions on a replica and another replica has all versions, then query returns a wrong result. If query may return a wrong result, it should be controlled per query and users should know that.

[dataroaring]

We'd better use a session variable to control behavior of query. Recovery is different than query, recovery is a action without lost, however, if query ignores some missing versions on a replica and another replica has all versions, then query returns a wrong result. If query may return a wrong result, it should be controlled per query and users should know that.

[xy720]

Ok, I understand. I will use a session variable to make this behavior only effect one session.

We'd better use a session variable to control behavior of query. Recovery is different than query, recovery is a action without lost, however, if query ignores some missing versions on a replica and another replica has all versions, then query returns a wrong result. If query may return a wrong result, it should be controlled per query and users should know that.

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[xy720]

run buildall

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[doris-robot]

TeamCity be ut coverage result:
Function Coverage: 37.04% (8301/22411)
Line Coverage: 29.20% (66663/228334)
Region Coverage: 27.83% (34611/124366)
Branch Coverage: 24.41% (17578/72002)
Coverage Report: http://coverage.selectdb-in.cc/coverage/f096b41aaf47ad5f30fbb86a151b0ac7963d841f_f096b41aaf47ad5f30fbb86a151b0ac7963d841f/report/index.html

[xy720]

run p0

[xy720]

run buildall

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[doris-robot]

TeamCity be ut coverage result:
Function Coverage: 37.05% (8303/22411)
Line Coverage: 29.21% (66701/228334)
Region Coverage: 27.85% (34630/124366)
Branch Coverage: 24.43% (17593/72002)
Coverage Report: http://coverage.selectdb-in.cc/coverage/578f00892830a5fb8c6ccd21faddcfc29cb8d506_578f00892830a5fb8c6ccd21faddcfc29cb8d506/report/index.html

[doris-robot]

(From new machine)TeamCity pipeline, clickbench performance test result:
the sum of best hot time: 45.95 seconds
stream load tsv: 552 seconds loaded 74807831229 Bytes, about 129 MB/s
stream load json: 20 seconds loaded 2358488459 Bytes, about 112 MB/s
stream load orc: 65 seconds loaded 1101869774 Bytes, about 16 MB/s
stream load parquet: 32 seconds loaded 861443392 Bytes, about 25 MB/s
insert into select: 28.9 seconds inserted 10000000 Rows, about 346K ops/s
storage size: 17161969847 Bytes

[xy720]

run p0

[xy720]

run p0

[xy720]

run buildall

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[github-actions]

clang-tidy review says "All clean, LGTM! 👍"

[xy720]

run buildall

[doris-robot]

TeamCity be ut coverage result:
Function Coverage: 37.42% (8487/22680)
Line Coverage: 29.81% (68915/231216)
Region Coverage: 28.27% (35670/126154)
Branch Coverage: 25.15% (18278/72666)
Coverage Report: http://coverage.selectdb-in.cc/coverage/36fad9a1e47a77fb5e2f1124dd96eb1c176ac507_36fad9a1e47a77fb5e2f1124dd96eb1c176ac507/report/index.html

[doris-robot]

(From new machine)TeamCity pipeline, clickbench performance test result:
the sum of best hot time: 46.04 seconds
stream load tsv: 559 seconds loaded 74807831229 Bytes, about 127 MB/s
stream load json: 20 seconds loaded 2358488459 Bytes, about 112 MB/s
stream load orc: 64 seconds loaded 1101869774 Bytes, about 16 MB/s
stream load parquet: 32 seconds loaded 861443392 Bytes, about 25 MB/s
insert into select: 29.1 seconds inserted 10000000 Rows, about 343K ops/s
storage size: 17162044251 Bytes

[dataroaring]

LGTM

[github-actions]

PR approved by at least one committer and no changes requested.

[github-actions]

PR approved by anyone and no changes requested.

[xy720]

run feut

[xy720]

run pipelinex_p0

[lide-reed]

LGTM