NETGEAR is aware of a growing number of phone and online scams. To learn how to stay safe click here.
Forum Discussion
Solved
RN524X RAM upgrade
I am going to upgrade the RAM on my RN524X and just wanted to pass along information as I go for anyone in the future seeking the same help, since this procedure will VOID your warranty, Netgear will...
Alright just wanted to report back again, the machine has been running great for a week with just the one good module installed, so all is good now, the memory was spec'd fine. The important point there was that one of the modules was/is faulty and the boot menu memory test did not spot that after running 25 hours with both modules installed. It did find come up more directly as a problem when running the boot menu memtest with just the one faulty module installed, the tester didn't spot it, but it crashed within 5 minutes after starting the test, two times in a row. And with that module installed (or both), the machine would boot up and run for a while, for hours..then spotaneously crash at some point, usually under some kind of CPU load...forcing a raid resync everytime, which was concerning to be honest, hopefully the data was not compromised, but anyway... seems to be sailing smooth now with just one 16gb module installed (see spec above). The faulty one has been sent back to HK for a refund. 16gb is more then enough anyway, I didn't really need 32gb.
So what performance differences do I notice after a week running smooth?
- a lot more caching memory is available and I notice that when I mount the raid from my mac over SMB, it is able to retrieve the folder contents list of certain folders with thousands of files, MUCH faster then before. This must be because that information is getting cached more now. However, i also notice that SMB process seems to run a little bit every once in a while to build that cache again when it needs to. But before I often had to wait minutes to open a folder and wait to see the conents (7000 files). Now it is often immediate.
- Before I had to setup a cron job to start crashplan and midnight and stop it in the morning because it was using nearly all of the factory 4gb memory. Now I can easily leave it running all day long and it barely causes the CPU to blink.
- Before gitlab was pretty much unusable with only 4gb of memory. Gitlab specs a need for 8gb+, and gitlab now runs perfectly fine and leaving it running all day, once it starts up, the CPU is having no problems.
- I am able to have up and running at all times gitlab+postgres, docker, portainer, crashplan, wordpress+MySQL, qdirStat, sabnzbd, sickchill and plex. No problems with memory, its using most of time only about 6gb of memory, and a nice large cache.
I do not think most people need to upgrade the memory on this box. I did so mainly to run gitlab, and also because crashplan was pushing it to the limit. What I can say is that if you have more advanced needs and are creatively using this box for stuff like I am, then its a pretty easy upgrade, once you locate the memory (not easy) and presuming the memory is not bad..which I might have just got unlucky that time.
Marking this thread as solved.
I don’t know if heat is the issue or not in this case. I removed one of the sodimms hoping that would increase airflow around the memory, but I can still get the machine to reboot after a while by loading up processes. Sometimes it reports panic and sometimes not. I am not able to be looking at the temp at the moment it happens but it seems to be running with door open and some decent load at around 65c which is not that unreasonable. There are no memory temp monitors so no idea about that. It’s weird to me that the memory reported no errors and can run for hours without problems and then after a while, perhaps when certain intensive things are happening the machine will panic or reboot out of nowhere.
Running out of ideas here now...
Running out of ideas here now...
Does anyone here think there is any reason that 2Rx8 SODIMM's would not work in the RN524X, as opposed to the 1Rx8 SODIMM that came in it?
I'm trying to decide if i should try out some other memory modules first before giving up on this. I am wondering if dual rank (ie, 2R), versus single rank (1R) might be a factor to why the box is panicing or rebooting for no reason and unpredictably. Either that or else 16gb per SODIMM is still some kind of problem. The ones I tried were 2Rx8 in 16gb per module. And like I said... No errors during memory test, seems to run fine on normal boot for hours at a time, then something casues it to occasionally have a problem and kernel panic sometimes or reboot abnormally, particularly under load.
Thinking about trying 8gb module size, but I can still only find that in 2Rx8. This is much more commonly found out there. I can't find any 1Rx8 at all now. Makes me wonder what Netgear is sourcing for currently sold RN524X units.
For DDR3 and above, it rarely matters. And it's also not normally the number of ranks that's a problem unless the board just can't handle that much (doesn't seem to be the issue). When it does matter, it's normally the data path width per chip (the X8 vs X4) that matters.
matters how?
Do you have any theories about why this SODIMM is not 100% working ok? It matches the specs I was given. Passes memory test, works fine for hours then fails under load...
Perhaps looking through the logs will pinpoint an error source. The only thing is that I don't know how to get the logs for the previous power cycle (where the problem occurred), just the current one (where there may be a clue, but it may be more subtle).
- I downloaded all the logs and some of them go back before the last reboot. But I haven’t been able to find anything so far just poking around in them.
If the problem is a kernel panic there probably won’t be anything there.
So I went to remove the one sodimm and go back to my original memory but I wanted to run a memory test again on the new sodimm first. This time the memory test on that sodimm did not go very long, less then five minutes and reboot without warning. It did that twice in a row. Do something is not right there, could not even run the test. Mind you it ran for nearly 24 hours booted up prior to that, so this is all very strange.
Hmm
So I swapped the modules and the memory test in that one is still running. It’s quite possible one of the modules was in fact bad. But I would have thought the memory tester would have reported errors rather then just rebooting the box spontaneously.
Stay tuned. In the meantime if anyone had any theories about these problems please advise The test itself must, of course, run out of RAM. So, it crashing isn't all that odd if the memory is bad.
well swapped the modules and so far it has passed 24 hours of memory testing and I booted it up and have been running pretty much all the stuff the loaded up the machine before...and so far....it has not restarted.. So.. hopefully that's all it is, one of the modules was maybe bad, even though somehow the memory test actually passed when both modules were installed.
So far so good. 16gb...running everything a little better. I
Alright just wanted to report back again, the machine has been running great for a week with just the one good module installed, so all is good now, the memory was spec'd fine. The important point there was that one of the modules was/is faulty and the boot menu memory test did not spot that after running 25 hours with both modules installed. It did find come up more directly as a problem when running the boot menu memtest with just the one faulty module installed, the tester didn't spot it, but it crashed within 5 minutes after starting the test, two times in a row. And with that module installed (or both), the machine would boot up and run for a while, for hours..then spotaneously crash at some point, usually under some kind of CPU load...forcing a raid resync everytime, which was concerning to be honest, hopefully the data was not compromised, but anyway... seems to be sailing smooth now with just one 16gb module installed (see spec above). The faulty one has been sent back to HK for a refund. 16gb is more then enough anyway, I didn't really need 32gb.
So what performance differences do I notice after a week running smooth?
- a lot more caching memory is available and I notice that when I mount the raid from my mac over SMB, it is able to retrieve the folder contents list of certain folders with thousands of files, MUCH faster then before. This must be because that information is getting cached more now. However, i also notice that SMB process seems to run a little bit every once in a while to build that cache again when it needs to. But before I often had to wait minutes to open a folder and wait to see the conents (7000 files). Now it is often immediate.
- Before I had to setup a cron job to start crashplan and midnight and stop it in the morning because it was using nearly all of the factory 4gb memory. Now I can easily leave it running all day long and it barely causes the CPU to blink.
- Before gitlab was pretty much unusable with only 4gb of memory. Gitlab specs a need for 8gb+, and gitlab now runs perfectly fine and leaving it running all day, once it starts up, the CPU is having no problems.
- I am able to have up and running at all times gitlab+postgres, docker, portainer, crashplan, wordpress+MySQL, qdirStat, sabnzbd, sickchill and plex. No problems with memory, its using most of time only about 6gb of memory, and a nice large cache.
I do not think most people need to upgrade the memory on this box. I did so mainly to run gitlab, and also because crashplan was pushing it to the limit. What I can say is that if you have more advanced needs and are creatively using this box for stuff like I am, then its a pretty easy upgrade, once you locate the memory (not easy) and presuming the memory is not bad..which I might have just got unlucky that time.
Marking this thread as solved.
NETGEAR Academy
Boost your skills with the Netgear Academy - Get trained, certified and stay ahead with the latest Netgear technology!
Join Us!
