Number of indexed reads differ from expected and depends (only in 2.5) on
presence of trace session
---------------------------------------------------------------------------------------------------
Key: CORE-4797
URL: http://tracker.firebirdsql.org/browse/CORE-4797
Project: Firebird Core
Issue Type: Bug
Components: Engine, TRACEMGR
Affects Versions: 3.0 Beta 2, 2.5.5
Reporter: Pavel Zotov
Priority: Minor
Following was done each run on empty database, page = 4k, FW = ON.
DDL:
====
create or alter view vmon as
select r.mon$record_seq_reads seq_reads,r.mon$record_idx_reads idx_reads
from mon$attachments a left join mon$record_stats r on
a.mon$stat_id=r.mon$stat_id
where a.mon$attachment_id = current_connection;
commit;
recreate table test(x int);
commit;
insert into test select rand()*100 from (select 1 i from rdb$types rows 100) a,
(select 1 i from rdb$types rows 100) b;
commit;
create index test_x on test(x);
commit; ------------------------------- [1]
TEST-1. Without trace.
======
select * from vmon; ----------------------- [a]
out nul; select min(x) from test; out;
commit;
select * from vmon; ----------------------- [b]
Result in 2.5:
~~~~~~~~~~
SEQ_READS IDX_READS
========= =====================
10173 382
SEQ_READS IDX_READS
========= =====================
10173 385
Difference of values in IDX_READS column is 385 - 382 = 3 rather than expected
1 (one).
The same result will be for: select x from test order by x rows 1;
Result in 3.0:
~~~~~~~~~~
SEQ_READS IDX_READS
========= =====================
10178 355
SEQ_READS IDX_READS
========= =====================
10178 357
Difference of values in IDX_READS column is 357 - 355 = 2 rather than expected
1 (one).
The same result will be for: select x from test order by x rows 1;
TEST-2. WITH trace launching after "commit; --------- [1]" and before first
subsequent query to view "vmon".
======
select * from vmon; ----------------------- [a] -- this should be issued AFTER
trace session will launched in another window
out nul; select min(x) from test; out;
commit;
select * from vmon; ----------------------- [b]
Result in 2.5:
~~~~~~~~~~
SEQ_READS IDX_READS
========== =====================
10173 382
SEQ_READS IDX_READS
========== =====================
10173 386
Difference of values in IDX_READS column is 386 - 382 = 4 rather than expected
1 (one). And this value differ from 3 that was received before, without trace.
Result in 3.0: the same as before, difference is 2 rather than 1. Trace has no
affect on this.
~~~~~~~~~~
Q1: why differences in IDX_READS column never was 1 ?
Q2: how TRACE session could affect on result of gathering statistics in
MON$RECORD_STATS in 2.5 ?!
PS. Tested on: WI-V2.5.5.26870 and WI-T3.0.0.31836, both - SuperClassic.
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators:
http://tracker.firebirdsql.org/secure/Administrators.jspa
-
For more information on JIRA, see: http://www.atlassian.com/software/jira
------------------------------------------------------------------------------
One dashboard for servers and applications across Physical-Virtual-Cloud
Widest out-of-the-box monitoring support with 50+ applications
Performance metrics, stats and reports that give you Actionable Insights
Deep dive visibility with transaction tracing using APM Insight.
http://ad.doubleclick.net/ddm/clk/290420510;117567292;y
Firebird-Devel mailing list, web interface at
https://lists.sourceforge.net/lists/listinfo/firebird-devel
