FYI: I’ve had the 100% usage bug too, and it went away by disabling miniupnpd on my openwrt. (which I don’t want to do, I depend on it 
)
Then again, having a server core constantly pegged at 100% also isn’t great. Would like to see this fixed.
First, Please make a conscientious sizing of that value. 2M worth, at 540 bytes each is a lot of non-swappable kernel memory to chew up (1+GB)
recommend you run a find /top/dir/one /top/dir/two/as/well -type d -print | wc -l will give you current values if you list all the TLD’s before the -type qualifier . Now round that up to the nearest power of 2 and you’re done untl you have that many directories monitored in real time.
As for this UPNP thing. I think the only way to figure out where it’s going is to wireshark capture it and actually examine the packet exchange. WIth the CPU running pegged, I bet your LAN is burning up too.
This does explain why I can’t reproduce it on either the ISP-provided modem or the pfSense’s upnpd implementation. The questions are: What is the actual traffic being exchanged and what should it be? What is being mishandled and where?
Could you assist me in capturing the required data for you?
If the CPU is constantly pegged, you would expect the LAN being very active as well indeed.
(though in the last month - before I noticed the issue - I didn’t experience any particular lan slowness)
(Ubuntu server behind OpenWRT router here)
Hi @ChuckPa, not wanting to hijack @heynnema’s thread, but last we discussed the UPnP problem you were quite clear to all involved in the various threads that you didn’t want anymore diagnostics as the issue had been raised with engineering and was out of your hands. Has this changed now, do you want more diagnostic info from those of us still suffering with this PMS bug?
Chuck,
I did "find ~/Pictures/ -type d -print | wc -l " and got 218, so it doesn’t sound like I’ve got too many folders in ~/Pictures causing my problem.
And… get this… I started PMS this morning, and it’s working normally. Not pegging the CPU at 100%! Nothing has changed.
Anything else I should check for now, or wait until/if it happens again.
Chuck,
I just reviewed the latest .zip file that I sent you, and I find a lot of curious network things… it’s trying to access some HDHomeRun tuners, that are currently powered off, at 192.168.0.201 and .202. It correctly finds .203. Also curious is it’s spending a lot of time trying to access a network extender at 192.168.0.10.
Chuck,
I just did some port scans of my network equipment, and my main router at 192.168.0.1, and a network extender at 192.168.0.3, both have open TCP port 5000. The main router has upnp turned off. The extender has no such option.
Important?
Port 5000 is SSDP IGD advertisement.
If the extender isn’t the IGD, it’s better to turn that off.
Chuck,
I found a recent folder of pictures in ~/Pictures with some .dot_filenames, ie: .file1, .file2, etc. Do you think that might have any impact? I deleted the .dot_filenames.
dotfiles should be ignored. They would only make the scanner skip over them that one time.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.
