Data corrupt?

[wypwong]

HI

I have this scernaio at work, some data for a particular company just gone missing. The period for the missing data has already been changed to read only, the only one can overwrite the data are admin users. We don't go into the production data unless we have to.

Is that possible that a restart of a server in the middle of day can cause the corruption? I have tried to look at the transaction log, but it's not giving me much info.

What else can cause the data disappear?

[jim wood]

If you didn't complete a save correctly before you shut the server down you may have lost data. Especially if logging for the cube was switched off. If you've made saves since the data was put in for that period that it has to be an update of some kind. May be somebody copied over the cub file while it was down with an older version?

[Carl Lonsdale]

The period for the missing data has already been changed to read only

I assume you have a Admin/Manager - Lock/Unlock Process for manual data entry and a specific period in which users have to enter said data?
If it is a larger department, one of the issues we have run into in the past is users stepping on each others toes. (Someone put their numbers in, only to have someone else zero them out a couple hours later) Although this typically is cleared up via Transaction logs.
Was the data locked possibly before the commit?
Also if you are using a sandbox/commit model for data entry there are some additional considerations and quirks that can pop up.
Lastly do you have a record of the data being there to start? I have on occasion run into the old: "I put in my numbers last night I swear!" Only to find out they had forgotten.

If there was any type of back-end TI processing in the near recent past, I would review that as well.
We have had holes appear in data especially around Qtr close processing and more prevalently around year end processing which required parameter switches, etc. Typically logging is disabled for these as well, which could as Jim mentioned explain why Transaction logs aren't helping.

Also check if logging is turned back on, if there is one sub-process towards the end of a series that forgets to re-enable it, it can be a pain in the butt.

[wypwong]

Sorry for the late reply, my inbox has put all the tm1forum emails to junk...

Carl, we are using TM1 as reporting, so we don't use the sandbox/committ model. And the data was defintely there, as I have copied the data from production to development about a month ago, and I can see the data in development.
Thanks on the point about TI process, I will double check on that.

According to the regional FC, this particular country always have data corruption problem (none of other countries have), I will sit down with the user and go through how they do their processes.