I setup my C7 to reboot automatically when the memory is too low. Usually it would last about a month before triggering a reboot. Since updating to 2.5.0 my hub was rebooting every day, so I lowered the threshold. But doing so makes my Zwave acting up and some lan devices not always responding to commands.
Seeing the sale on C8 pro, I thought perhaps it was time to upgrade. But residing in Canada and with the trading situation, I'm not sure what it will end up costing me.
Anyone has an idea of how much extra fees this purchase could amount to?
step 1 for me was removing all unneccessary apps, rules and chatty devices. Oh and LAN integrations were hell for me.
EDIT - just updated my C7, the free memory from a clean boot was.
2.4.4- 638k before upgrade
2.5.0.146- 424k after upgrade
So you're absolutely correct, something is up. Maybe it will stabilize better long term?
And yes a C8pro makes this much better, but IMHO that's throwing hardware at a long standing memory leak issue, which might be causing other issues?
I would love to let you know, but I have not purchased much directly from the US in a while, now that Amazon Canada has more and more stuff, it's often not worth the hassle for the few dollars you save. I know Aartech has the C8 Pro in stock at 260 CAD + taxes which is very reasonable based on normal pricing but the sale price of 219 CAD is pretty sweet if you don't get hit with import fees, it's hard to say.
All my 3 hubs were bought and shipped in the US (sister lives in the US), so I did not have to worry about the import fees. Maybe someone else can chime in on import fees?
Same here on my C5. Right now I'm at 105K free and surprising, things are still functioning but I have seen z wave "get funky" when memory gets low. I had the floor at 160K but disabled that because I'm hitting that in a day or less. I did restore the prior version and ran that for a week and everything was back to normal so it's definitely 2.5.0.
The C5 is has been rock solid so I have resisted upgrading. Why mess up something that works. I also have quite a few devices that have followed me since I started this in 2015 and that is concerning. I know, hub mesh but that is adding complexity. I hoping that if enough people speak up maybe staff can look at the hubs and see what changed for some of us......
It is not just you (and I have noticed it as well). I have been having to reboot as well nearly every other day.
100% this. Something is amiss even if it does appear to be only anecdotal at this point.
Not really wanting to spend money on the upgrade to fix an issue that did not exist prior to the update. Especially when I have to go and update a bunch of links on my and my wife's phones in tasker for some automations we both have that use cloud links as a fall back for when VPN might not work.
You can buy From aartech.ca and don't have the worry about all this tariffs stuffs. I ordered quite a bit of stuffs from them and quite happy with their service.
Amazon.ca for next day delivery and crazy return policy but a little more pricey.
Or take your chance direct from Hubitat. Even with duty and tariffs it will probably be similar to Amazon at the end I would think.
Immediately after the upgrade to 2.5.0 I also had the same issues on my C8, with my low Free Memory rule leading to once and twice daily reboots.
When I saw that things stull worked ok at lower Free Memory levels, I set the triggers to 40MB for immediate reboot and 50MB for my nightly reboot check. Then I updated to the latest release (2.5.0.143) and it went for a full day without rebooting. The Free Menory is extraodinarily low based on previous history, but doesn't go down as rapidly as on 2.4. Here is a 14 day Watchtower chart of freeMem, dbSize and uptime in minutes. This shows the big difference between 2.4 and 2.5.0
Thus next chart is for 7 days to see how FreeMem went down in 2.5.0.136 (installed 5/16 5:30pm) and then stabilized in 2.5.0.143 (installed 5/20 7:51am)
Finally let 2.5.0.143 run without a reboot and after the first nightly system cleanup routine it seems to be running well with very snappy response to various triggers. This last chart is for 2 days. It shows the nightly cleanup and several system cleanups in between.
I'm going to let it run to see what happens after a week. May be able to remove the database rebuild reboot rule that runs it after 7 days or at least extend the time.
Fingers crossed that the snappiness stays there.
Not sure if you need to replace your hubs yet. Just let them stabilize on the new version for a few days. C7 hubs may have enough memory to stabilize.
I have found that with shipping the total price comes too close to Aartech’s to bother with that risk. Wish they would honor promotional pricing though.
I concur - AArtec is the low risk approach. Duties and brokerage are hit and miss with USPS, but guaranteed with the other legacy shippers (UPS, FedEx, ETC.)
I think the title of the tread is a little mis leading. A memory leak isn't something you would see right after a reboot, but over time. What you are seeing is simply a need for more memory based on all the features being added.
That said i do believe there are allot of new features that have been getting added that just simply need memory. I certainly wouldn't recommend getting anything other the a C8 Pro.
That said it is hard to know without a better understanding of what you actually are running and have connected to the hub.
I admit I was pleasantly surprised with my Ring.com order's free shipping last Oct.
It was shipped UPS International, some sort of tier that had no extra charges/fees.
I have an 8 Pro so the memory leak doesn't affect me as much, but I have say that there is a pretty bad memory leak. I've seen it go from 947mb to 135mb free memory inside of a week. If there's anything we users can do to help the Hubitat developers track this down, please tell us.
It isn't as much a memory leak as it is simply a higher need for memory.
The nature of how java works will reclaim memory as needed based on them being expired out of memory. That is based on memory presure and garbage collection cleanup configuration. If the memory presure is not not there it may not reclaim items from memory. Keeping things in memory is a good thing. So you also don't want to be to agressive with removing items from memory. It is a good thing to use the memory if you have it.
I could replace the title with "Higher need for memory = time to upgrade?", but the question remain the same.
While reading your explanation I wondered: Are you suggesting the C7 was lacking memory right from the start? Because this decreasing amount of available memory has always been there. That is why I had this rule about rebooting on low memory. It was manageable because it would take about a months before the threshold was reached. Now, to get the same result with that "elevated memory need", I have to lower the threshold to a level that render my hub noticeably less reliable.
Same here. It could take a month or two before it rebooted. It would reboot faster if I was working on rules or dashboards within the hub. The memory leak is much slower if I left the hub alone and didn't change anything. It seems to do the same thing with 2.5.0. The memory decreases substantially faster (which makes sense because you are doing something) but doesn't come completely back after the cleanup process.
The hub may now need more memory but it sure looks like it's leaking some memory too. If this is the case though that 2.5.0 requires this much more memory then there are C5s (and it looks like some C7s too) that have come to the end of the line for upgrades.
It would be great to get a comment from someone at HE. I'll tag Bobby and see if he will comment next week. @bobbyD