Thanks Jeffrey. Very helpful. Our production server has 168 GB of RAM. Our app uses ~68 and another ~5 is consumed by OS overhead etc. We generally run with ~95 GB of Ram free, so I guess we’ll be okay…
I’ll run a test convert after hours on the production server and see if it completes without a problem. We do reboot our 4D app once per week. (Hold-over habits from past years of less stable 4D products.) You mentioned this hang behavior sometimes appears when quitting 4D Server. Have you seen it and does it recover (or how long does it hang)? I will file a ticket with 4D. Thanks again! Kirk > On Apr 12, 2018, at 9:49 AM, Jeffrey Kain via 4D_Tech <4d_tech@lists.4d.com> > wrote: > > Kirk, > > According to 4D, the problem is due to the new cache design in v16. We've > also seen this on our servers that have a large amount of memory installed. > We're working with 4D on a solution to this. The problem, according to 4D, is > that the new cache manager allocates a lot of very small blocks of memory, as > opposed to always four very large blocks in v15 and earlier. When it comes > time to release some of the memory when cache is full, something goes wrong > and Windows takes a very long time to release all of these tiny blocks. > > Three solutions: > > 1. Wait it out. If you open task manager you can watch the amount of memory > used by 4D slowly fall. It will eventually come back to life and continue. > > 2. Add enough memory so that a cache purge is never necessary. > > 3. Allocate a fixed size cache that will handle the entire re-indexing > operation even though it will result in page file swapping. If you're on an > SSD, the swapping isn't so bad and will allow 4D to complete its indexing. > Once this is done, set your cache back to a normal value for your application. > > Note that the same problem will happen in normal use if you fill the cache > and 4D needs to purge some objects. This also happens sometimes when you try > to quit 4D Server. 4D has recommended that we set the percent of cache purge > to a small value in the database parameters, but I haven't tried this yet. > We bit the bullet and installed enough memory in our server so that a cache > purge will never happen. > > I know 4D is working with Microsoft on this. You might open a tech support > case so your voice can be added to ours. > > -- > Jeffrey Kain > jeffrey.k...@gmail.com > > > > >> On Apr 12, 2018, at 12:30 PM, rooftop99--- via 4D_Tech >> <4d_tech@lists.4d.com> wrote: >> >> The pattern is as follows: >> -Open v15 database with v16R5 and authorize the conversion. >> -Various Building Index and progress bar dialogs appear >> -Task Manager shows very little CPU activity, but the Memory used graph >> slowly increases until it pegs at 99% of physical Memory installed on the >> machine. >> -Eventually the 4D dialogs stop updating and the machine bogs down to near >> non-responsive. (I have left is in this state for hours to see if it would >> recover, with no luck.) >> -I have to kill the 4D app and reboot the server. >> -Once rebooted I can open the app again with v16R5 and it appears to start >> where is left off and build the rest of the indexes, etc >> -Eventually the 4D update process completes and the app appears fine... > > ********************************************************************** > 4D Internet Users Group (4D iNUG) > FAQ: http://lists.4d.com/faqnug.html > Archive: http://lists.4d.com/archives.html > Options: https://lists.4d.com/mailman/options/4d_tech > Unsub: mailto:4d_tech-unsubscr...@lists.4d.com > ********************************************************************** ********************************************************************** 4D Internet Users Group (4D iNUG) FAQ: http://lists.4d.com/faqnug.html Archive: http://lists.4d.com/archives.html Options: https://lists.4d.com/mailman/options/4d_tech Unsub: mailto:4d_tech-unsubscr...@lists.4d.com **********************************************************************
