FMS 15v3 huge issue : Client-Server Cache gets corrupt / out of sync, SQL Queries unreliable

Discussion created by Vincent_L on Nov 29, 2016
Latest reply on Jun 19, 2018 by Vincent_L



Last week, when I installed 15v3, a query that always worked, failed.

What was strange is that the backup, launched standalone (not hosted) worked.

But  I spend a day rebuilding all my indexes, and then it seemed ok


Now, a week later, same issue appears again. So I backuped the file, made a copy of it, renamed it, and hosted the copy.

The copy had no issue. Then I tried something different. I installed a FMPa 15v2 client ON THE SERVER machine (mac os x sierra 10.2.0). And launched the query on the original hosted file : it worked.

Then I launched on my main mac FMPa 15v2, exactly same query and got the question mark ?


So, same query doesn't work the same on 2 different clients. Fearing a dns issue, I made sure to open the file with the same fmnet path

I also removed plugins


Then I tried FMPa 14 on may main mac : it works.



FMPa 15v32client on same mac as server : works

FMPa 15v2 on main mac : fails

FMPa 14 on main mac : WORKS

FMP 15v2 on another worker mac : fails


So, to me, I'm pretty sure that FMS 15v3 introduced a bug in the caching mechanism introduced in the Filemaker 15 platform.

Or that the existing cache created on clients with FMS 15v2, don't work with those created with FMS 15v3

Or more probably (as it worked on my main machine after I installed and rebuilt the index of all my file), that slowly with FMS 15v3 and 15 cleints, the cache between the 2 gets corrupt or out of sync.


I must say that if I limit the items of the query, it works on both. So this leads me to think, that only part of the cache is corrupt or out of sync.


To validate this I should delete my main mac client cache, but I don't know where it is


N.B : Just confirmed the cache corruption hypothesis, on my same mac I created another user and tried with FMPa15v2, exact same query and it worked


P.S : I'm keeping my mac that state today, so if you want me to zip the cache files and send them to you along the main file, tell me quick


P.S.2: the query is the one below, last week test showed that removing the JOIN category C ON  C.catid=I.Catid, or doing this

JOIN category C ON  C.catid=5586 worked. But I think it's not the reason of the issue (but maybe JOINs are more sensitive to the issue).


ExecuteSQL ( "SELECT DISTINCT E.Product_ID  FROM Products_EANs E JOIN files_index I ON  I.Product_ID= E.Product_ID JOIN category C ON  C.catid=I.Catid WHERE E.EAN IN ('0045496334420','0045496401795','0045496473877','0045496522513','0045496523893','0045496526702','0045496527471','0045496528607','0093155148963','0734646967136','0734646257961','0734646964517','3298060230506','3660116051234','3165140465304','3165140582667','3165140658799','3165140649544','3165140711678','3165140579087','0711719221142','0711719825227') ORDER BY FMP_arborescence_name";"";"" )