[ROLLBACK] - inserted value too large for column: 9000000 >>

[Former Member]

has anyone seen this yet?

i'm loading 523362 into a column table and exactly 5000 are written with this message:

[ROLLBACK] - inserted value too large for column: 9000000 >>

so HANA shows holding only 518362 and error file is exactly 500KB. i have checked the dumped records and they seem to be similar to those that do load.

any suggestions?

[Former Member]

bad record:

GM02,2005,100042998,10,134001,-24.22,H

good record:

GM02,2005,100040144,199,220002,-1332788.08,H

dictionary definition:

char (4), char (4), char (10), char (3), char (6), decimal (16), char (1)

[tomas-krojzl]

Hello,

I noticed interesting point - data type CHAR is not mentioned in guides anymore... Questions is - was it deprecated?

I would suggest to use VARCHAR for ASCII characters (0-128) and NVARCHAR for Unicode strings.

Tomas

[Former Member]

changing from CHAR to VARCHAR results in rejecting all (100%) records with this error message:

[ROLLBACK] - inserted value too large for column: 10003

i'm referring to SQL Data Types as i'm not sure how to change Column Store Data Type.

[tomas-krojzl]

Hello,

please provide me all required information so that I can reproduce the problem...

- SQL you used to create the table

- sample of data (10 correct lines + 10 problematic lines)

- command you use to load the data

- all other files...

also please what revision are you using?

Tomas

[lbreddemann]

Hey Tomas,

want to join SAP support ?

In fact, apart from posting all this stuff into this thread, Gregory could (and maybe should) just open a support message.

We'll then not only have the option to try and reproduce the behaviour on our systems, but also to logon directly to his HANA box and analyse the issue with the version used here.

Best regards,

Lars

[tomas-krojzl]

Hello,

want to join SAP support ? 

Heh.. thanks for offer...

In fact, apart from posting all this stuff into this thread, Gregory could (and maybe should) just open a support message.
We'll then not only have the option to try and reproduce the behaviour on our systems, but also to logon directly to his HANA box and analyse the issue with the version used here.

Sometimes it is much faster to go to SDN rather then to use support. (this is not critics it is fact)

Some situations are easy to solve and community can help much faster and better then SAP support team. On the other hand sometimes when the problem is more related to internal logic (which might not be yet published) - SAP support team is the only one who can help. I think that this is very nice symbiosis where we (guys on SDN) act as first barrier removing "easy" problems from your shoulders so that you can focus on solving the tough ones.

When we are with SAP messages I opened message (1007743 / 2011) for the SUM issues described here:

I think this is nice example of issue that only you can solve as there is not even one hit anywhere regarding the SAPverifytest.pse that seems to be required for this scenario.

(And I am going to open second SAP message because after applying SLT SP05 the SLT is not working at all... But this one I want to test again one second system without any SAP notes applied.)

Tomas

[Former Member]

Hi Tomas,

thank you for the great answers you have given me and the rest of the SDN forum participants. i will open a customer (SDN) message once i test my current hypothesis (maybe even a workaround). at the moment, this particular HANA system instance seems very logical in its rejections of records, but i need some time to fully test it out.

i'll be back

thx, greg

[lbreddemann]

Hi Tomas,

you're right - sometimes the community is a better and quicker place to seek support.

Actually that's one of the major reasons the forums have been put in place.

Having worked in SAP Support and in the SDN now for some years, I usually would recommend to place How-To questions to the forum and issues that smell like software bugs in support calls.

One point for that is: support calls can be easily led to a bug fix. There's a process behind that.

For forum threads, there is no regular development contact working on them...

Anyhow, thanks for being that active here - keep on doing that!

Cheers and a nice weekend,

Lars

(GSC Austria/Vienna - SAP Active Global Support - Primary Support)

[tomas-krojzl]

Hello,

One point for that is: support calls can be easily led to a bug fix. There's a process behind that.
For forum threads, there is no regular development contact working on them...

Maybe it would be worth to create some process - one employee responsible for monitoring the forums and when it comes to conclusion that something is bug - he might intervene, then try it internally and if confirmed he can raise internal bug request.

Reason for this is that sometimes it is difficult to separate bug from feature and it is faster to use workaround rather than use time consuming process of opening SAP message with unclear result. Example can be DUAL table - it is faster to use DUMMY instead and go forward rather then to complain that DUAL is gone and pushing SAP to "fix". Btw. is this bug or feature? 😄

Tomas

[Former Member]

Tomas, very well put. I think it is particularly relevant for products such as SAP HANA as it is early days.

From all indications, SAP is already at it. SAP's response on HANA has been fantastic and I can see they have spent lot on organizing the community and collaborative working.

A big Thumbs Up! to SAP I would say.

[Vitaliy-R]

I had similar issue in the past, but it was on Rev 6 or so.

[[ROLLBACK]] - inserted value too large for column: 2005000835636, >> 2005000835636,2,P,2300,1,1082467,u2026

Although target column for 2005000835636 was VARCHAR(13), so should fit.

At that time resolved by changing column type to

serialno CHAR(13) NOT NULL,

but it did not make sense for me.

I have a feeling that this may have something in common with the dictionary compression, but I do not know...

[tomas-krojzl]

Hello,

aha we are speaking about strings? Are you inserting Unicode non standard characters?

If answer to both questions is yes that it might be related to VARCHAR x NVARCHAR issue. VARCHAR is having length in bytes while NVARCHAR is having length in characters. In case you are inserting special chars - then you might have issue because these special chars are stored as multiple bytes and you will quickly run out of column size with VARCHAR even when it might not be apparent from the first sight.

Tomas

[tomas-krojzl]

Hello,

can you provide table definition, few lines that were loaded and few problematic lines?

Tomas

[tomas-krojzl]

Hello,

also please try to look in following directory:

/usr/sap/<SID>/HDB<system number>/<hostname>/trace

There you should have trace files - try to locate the error and you should have there more details about what went wrong.

(my tip are indexserver_* files - but better grep everything for the error message)

Tomas

[Former Member]

Hi All,

I don't find SID in this directory, Will hana support SID like oracle ... Please suggest me ...

Regards,

manoj.

[tomas-krojzl]

Hello,

I don't find SID in this directory, Will hana support SID like oracle ... Please suggest me ...

Which directory? SAP HANA database does have SID. And following directory should exist:

/usr/sap/<SID>

Tomas