My gut says the same, I have even experienced it practically while still in ROS 6 - I haven’t noticed those complaints but after a reboot, everything was running fine except that I have lost a few ppp secrets I have added over last several weeks. Since I am using the command line almost exclusively, I figure the complaints were simply not output to the console in 6.49.13.
I’m waiting for Mikrotik to respond to my ticket again after I have updated it with a bunch of supout.rif taken after each bite on a machine where the ghost started “eating” and returning the disk space, hoping that they will identify the process that is responsible for that and eventually not only fix the root cause but also add a cleanup .npk like they did back in the days (but that one was removing one particular file so even if it still worked, it would not resolve the current issue), so that devices could be fixed without physically touching them.
since 7.16-7.17, there seems to be some kind of problem that causes the size of the configuration to grow when changing.
my script periodically updates (but does not add) 25 entries in the /ip/dns/static section. each update spends 4K of free memory
I’m restoring my configuration a month before the problems (netinstall + backup restore), I have about 300KB free with the installed qc-ac, but then they leak
Technical support is not responding to my requests.
It doesn’t make any sense. until the problem is fixed, this will return the device to a working state here and now, but after a while it will happen again.
it makes no sense to argue that 16M is not enough: the configuration takes ~ 150 KB, and 300-600KB is free on the device with wifi-qcom-ac installed. What does little mean? It can accommodate 3 more such configurations, which is not so small.
Or do you mean too little for mistakes to go unnoticed? - Yes, but it will take time and devices with a lot of memory will suffer, it’s just a matter of time.
I am freaking out. I just found this thread because I have a Chateau 5G (the discontinued model) with 16 MB of storage. The device is a two-hour plane ride away from me and won’t reboot, showing the same error:
system,error,critical could not save configuration changes, not enough storage space available.
I have created an incident at Mikrotik SUP-180945 regarding storage place.
Below the answer:
From: Serhii T. (Jira) [mailto:support@mikrotik.com]
Sent: Monday, March 3, 2025 2:17 PM
To:
Subject: MikroTik support #[SUP-180945]: Unable to upgrade to RBD52G-5HacD2HnD
—-—-—-—
Please REPLY ABOVE THIS LINE ^ (for faster response, use our support portal).
Hello,
If the device has 0KB of free space during changes, the settings will not be saved, which is the expected behavior.
In this case, you can only reinstall the system on the device using Netinstall.
As the software evolves and consumes more disk space, it continues to fit within the internal storage capacity of the router.
We continuously improve and optimize RouterOS and its features to support our devices.
Consider reviewing your configuration to reduce space usage or upgrading to a more powerful device.
Best regards,
Serhii T.
Your eyes don’t fool You !
Advice from Mikrotik is to replace device or go for netinstall. Ridiculous !!! I have some of the devices in remote locations of my customers. Shall I spend time and money to travel from one to another ?
Is this Mikrotik quality ?
I have the same problem with a WAPAC where the disk space slowly disappears.. I removed a backup, no change to the reboot/configuration save issue. I then ran a supout and it wrote 350KB to the disk… So this is a OS calibration issue with usage, or there is a partition within the OS itself which has run out of disk space.
I have a similar script (turning dhcp leases to dns static entries).
I postponed the update quite a while and updated from 7.12.1 to 7.18.1 and now facing this 0 hdd error.
Does your rollback solved this issue? Did you tested on any newer version?
Still frustrating that nothing can tell what occupies the hdd space, so we could figure it out what can causes the loss.
Netinstalling 7.16.2 (using netinstall from 7.18) to hAP ac^2 gives ~660 KiB of free space. Netinstalling 7.18 leaves only ~100 KiB which drops down to 0 after some time
yes, rolling back to 7.15.3 solves the problem. if you don’t need the features of later versions, you can stay on 7.15.3 with the wave2 package installed. in my experience with several dozen different mikrotik devices, this is a fairly stable release.
Your CRS is one of those “16MB flash, 64MB+ RAM” which have root of file storage on RAM disk … and remainder of flash is mounted under /flash/ . Backups (and exports) are by default created on root of file storage, on your device that means on RAM disk. Your device has no shortage of RAM, so writing to RAM disk is perfectly possible.
I tested this rollback and it seems much better, but it still has some “HDD” eating issues. I made a couple changes in my scripts and they not touching the config file as much as they did before with the static DNS entries (so a bit ruined the test for the rollback, but this is my main router at home and need to be working).
On the left side there is a 24 hour graph showing the garbage collections.
On he right side there s a daily min/max graph showing that the max is getting lower (around 4.5kb) each day.
It would be so nice to be able to check what occupies this space, but if I see it right, there is no such tool.
If it would be saved on RAM the files would be gone when the system reboots. But they survived “hard” power off (removing power cable as I can not shutdown/reboot the system anymore)
Bricked one of my hAP AC2s due to this bug when trying to import a DoH root cert, apparently that pushed it over the edge and they cannot be deleted. And the shitty configuration import after netinstalling constantly errors and has to be handheld all the way due to conflicts with default configurations, wifi1 and wifi2 swapped radios for some reason, my wireguard keys were somehow lost, a terrible experience.
Would be nice if you would explain this. I am sure there are no secret batteries in the device to keep the data in the RAM when power is gone.
Power Off → all RAM is gone.
As I explained, and output of your files confirms it, that the remaining (which is 0) of flash is mounted under flash/ … and if you create any file to root (i.e. not under flash/), that file is in RAM disk. And RAM disk does get wiped at reboot (or power off).
And the fact is that 16MB flash is very tight for recent ROS versions. It’s possible that those 650kB worth of files can fit on flash (but not on root) after you install such device … but it is well possible that flash gets depleted right because of creating backups and exports.
So I have hard time believing you that those files survived power down of device. And based on your forum metadata (date of joining, number of posts) I have my doubts about your ROS experiences (but I may be wrong here).
The backup files were still present after the device was power cycled, so they must be stored on flash.
However, I received an explanation of what went wrong with the flash drive, and it seems that only Netinstall can properly clean it. The issue started after I ran a packet capture (pcap) without setting a size limit, which caused the flash storage to fill up. I would expect a professional system to clean up the filesystem on boot to prevent issues like the one we’re experiencing. The downtime required for Netinstall is significantly higher than just performing a reboot.
Regarding my background: I’m rarely online here, and until now, aside from some configuration issues, my MikroTik routers have worked fine in my network. The one I’m having problems with was purchased seven years ago and has done its job reliably. At work, we use enterprise-grade equipment like Cisco, Juniper, and others. I’ve been qualified to work with such systems for over 20 years.
A problem like the one I encountered here wouldn’t be an issue on the systems we use professionally.
Sorry to say this, but your responses were not helpful. Personally, I would have preferred no reply over the kind of answer I received.
My redponse to @OP was to netinstall device. You later chimed in with claim that there’s bug in flash handling in ROS (etc.) about which I expressed my doubts. Otherwise you (at least implicitly) referenced to @OP’s post (by posting in this thread).
Now you’re telling that MT support told you to netinstall device. So how is my suggestion to @OP different ftom what MT support told you?
And I still have my doubts about your explanation. Your extensive experience with equipment by other vendors don’t count much when dealing with ROS-specific problems.
And, for the record, I completely agree that ROS should be much more robust when it comes to handling out-of-storage situations, the way things are currently is completely unacceptable.
