what could i explore to create a reproducible example for a bug report?

yellow highlight is the only difference
# this command works fine create table zp13 as select v.* , arf.fips , arf.county , arf.countyst , arf.msacode from v13 as v left join arf on v.state_code = arf.state_cd AND v.bene_county_cd = arf.cnty_cd WITH DATA;
# this command crashes mserver create table zp13 as select v.* , arf.fips , arf.county , arf.countyst , arf.msacode , arf.msaname from v13 as v left join arf on v.state_code = arf.state_cd AND v.bene_county_cd = arf.cnty_cd WITH DATA;
i've attached the sys.columns for the two tables, not sure if attachments are allowed on the mailing list but i can send separately if needed. thanks!

hello, when i changed the column types from VARCHAR(255) and used only CLOB
types, the merge appears to work. so i believe the server dying must be
related to VARCHAR(255)=CLOB within two merge fields. any idea why that
might trigger a server crash? thanks
On Mon, Nov 16, 2015 at 5:00 AM, Anthony Damico
yellow highlight is the only difference
# this command works fine create table zp13 as select v.* , arf.fips , arf.county , arf.countyst , arf.msacode from v13 as v left join arf on v.state_code = arf.state_cd AND v.bene_county_cd = arf.cnty_cd WITH DATA;
# this command crashes mserver create table zp13 as select v.* , arf.fips , arf.county , arf.countyst , arf.msacode , arf.msaname from v13 as v left join arf on v.state_code = arf.state_cd AND v.bene_county_cd = arf.cnty_cd WITH DATA;
i've attached the sys.columns for the two tables, not sure if attachments are allowed on the mailing list but i can send separately if needed.
thanks!

hello, now my code is breaking the third time table zp13 gets created.
this code worked on prior versions of mserver. i am doing my best to
figure out the root, but it's in the middle of a multiple-day preparation
script.
when i do make it to the database corruption, this line kills the server:
select tables.name, columns.name, location from tables inner join columns
on tables.id=columns.table_id left join storage on tables.name=storage.table
and columns.name=storage.column where location is null;
i think that means that it is the same class of bug as this one that was
introduced by recent updates. was the fix for 3828 a genuine fix or just a
patch?
https://www.monetdb.org/bugzilla/show_bug.cgi?id=3828
is there anything that someone might have missed when fixing bug 3828?
thanks
On Mon, Nov 16, 2015 at 12:40 PM, Anthony Damico
hello, when i changed the column types from VARCHAR(255) and used only CLOB types, the merge appears to work. so i believe the server dying must be related to VARCHAR(255)=CLOB within two merge fields. any idea why that might trigger a server crash? thanks
On Mon, Nov 16, 2015 at 5:00 AM, Anthony Damico
wrote: yellow highlight is the only difference
# this command works fine create table zp13 as select v.* , arf.fips , arf.county , arf.countyst , arf.msacode from v13 as v left join arf on v.state_code = arf.state_cd AND v.bene_county_cd = arf.cnty_cd WITH DATA;
# this command crashes mserver create table zp13 as select v.* , arf.fips , arf.county , arf.countyst , arf.msacode , arf.msaname from v13 as v left join arf on v.state_code = arf.state_cd AND v.bene_county_cd = arf.cnty_cd WITH DATA;
i've attached the sys.columns for the two tables, not sure if attachments are allowed on the mailing list but i can send separately if needed.
thanks!
participants (1)
-
Anthony Damico