TimeMachine not working with wireless bridge setup

Benihana · 13 Oct 2014 at 13:26

So I have a Synology DS413j running as a TimeMachine server, backing up my rMBP and my girlfriend's MBP. It all works fine when cabled directly in to the main DSL modem (Zyxel VMG8924-B10A) however, the usual setup is to have it running off a wifi bridge. I have been using a Netgear WN3000RP for this. All was well until recently it has stopped 'finding' the TimeMachine, I get the error:

Time Machine couldn’t complete the backup to “DiskStation”. The backup disk could not be found. Make sure the backup disk is connected or select a different backup disk.

My thinking is there is a wireless setting, or networking feature I haven't turned on or off which is causing the issue. I have DHCP on for the Zyxel, and off for the Netgear as all devices attached to it have static IPs. I can browse all the shares, including TM backups from both MBPs and have no other issues than this one. I have AFP,SMB and NFS enabled.

Any ideas why this is happening? As I said, hardwired into the Zyxel there is no issue so it must be something with the extender or bridge connection right?

Cenedd · 13 Oct 2014 at 13:55

I've had similar issues when trying to use a TP Link box as a wireless client. After using Wireshark to see what was happening, I found that ARP was being mangled as it traversed the bridge and meant that hosts on one side couldn't find the MAC address of hosts on the other side that they needed to communicate with. You might find that a firmware update of the wireless access point in client mode may fix this or a recent firmware update may have caused it. I ended up switching to powerline Ethernet which worked for this location but because it was a large multi-storey officeblock, I only got 40Mbps bandwidth between the two points. Your environment may be more forgiving.

Benihana · 13 Oct 2014 at 14:06

Thanks for the info, NAS, Modem, Extender and MBP are all as up-to-date as they can be, I cant think of anything that happened around the time it crapped out. Just seemed to come out of nowhere. I'm downloading wireshark at the moment to see if I can confirm the ARP problem, I'd like to avoid powerline if possible, but it looks like it might be that or a new AC wifi bridge which will probably cost more..

Cenedd · 13 Oct 2014 at 14:11

I think I could get around it by putting static ARP entries on machines so they could find each other...but it wasn't manageable for the size of network I was working on. For a few machines it might work though.

Benihana · 13 Oct 2014 at 15:51

OK, looks like the issue is as you say, ARP related (target mac address is blank). I've tried adding a static arp entry on my MBP, rebooting all devices and still have the same issue. Perhaps I'm missing a step?

Benihana · 13 Oct 2014 at 15:57

Also enabled this option on the synology:

"Reply to arp requests if the target IP address is a local address configured on the incoming interface."

No change..

Cenedd · 13 Oct 2014 at 16:27

The target for the ARP Request should be all F's which is the broadcast MAC address - ie all machines in the broadcast domain. You should see something like
"who has 192.168.1.50? Tell <requester's MAC address>"
If I remember correctly the WAP (Wireless Access Point) in Client Mode did something like substitute it's own MAC so that the reply was sent to the WAP and not the original requestor - in this case, your MBP (Mac Book Pro).
If ARP is your problem, you won't be able to ping the IP address of the Synology box (Time Machine server) from your MBP if they're at opposite sites of the bridge because the MBP won't be able to resolve the Synology's MAC address. If the Synology allows ping tests, it's worth checking the reverse too. If you can ping both ways - and try rebooting between checks - then ARP is not your problem and we're barking up the wrong tree.

Assuming for a minute that the problem IS arp, do you have IPv6 set up? Just thinking that IPv6 doesn't use ARP, it uses neighbour solicitation which is ICMP and doesn't use broadcasts. Could be worth trying everything with IPv6 enabled and see if it springs to life - or is at least ping'able.

Benihana · 13 Oct 2014 at 16:48

Hmm might be looking in the wrong place then!

I saw this, looked like it wasn't getting the target MAC address.

I'm now getting a load of ARP entries of the Who has *ip address of extender*? tell *ip address of modem*...

I'll switch things to IPv6, I think I avoided this as it messed with AirPlay. I'm also not sure the extender has IPv6, but as its not DHCPing does this matter? Will post back when I've switched everything.

Cenedd · 13 Oct 2014 at 17:00

Try pinging the NAS from a MBP and see what you get.
Try the reverse if possible - or put a MBP at each end and try.

The target MAC will be blank for the ARP request as that's the information it's requesting.
If you ping from MBP to NAS, you ought to get a Request from MBP to IP of NAS with blank target MAC and that should be replied to by an ARP reply with the details of the NAS.
You may need to use wireshark at each end of the bridge to see what's getting sent vs what gets over the bridge and whether there are any replies that don't make it back.

Benihana · 13 Oct 2014 at 17:16

I've always been able to access the NAS interface and shared files, its usually pinged fine, with the odd dropped packet. I can't see an option to ping back, but not sure if it would show any different results?

I've gone through and switched things to IPv6, no joy, but then on switching them back I seem to be getting further than I did before. The MBP is now "preparing backup" so perhaps it is something to do with restarting the protocols on one of the devices..

Cenedd · 13 Oct 2014 at 17:28

If you can ping, the problem isn't ARP. Dropped packets are probably WiFi being Wifi. You shouldn't notice a major problem with less than about 5% dropped packets.

Sometimes setting it up again will fix it...but it's still frustrating not to know why. Hope it works for you.

Benihana · 13 Oct 2014 at 19:36

One step forward, two steps back. Pings are now yo-yoing, having to reboot devices left and right. When the pings are working the backup still isnt being found, yet I can browse the folder as an NFS share.

As it works cabled, i'm probably going to look at the powerline solution again. The Devolo 1200 units look quite tasty..

Cenedd · 13 Oct 2014 at 20:01

The powerline is probably going to be your simplest answer as this is what they were designed for rather than adapted to. That said, if it works cabled and pings work then it ought to work. There aren't any differences in say DNS servers when cabled or bridged are there?
Could it be something like MTU size issues? NASs often have settings for jumbo frames and they may not get across the bridge. Worth a quick go at turning off Jumbo frames or setting MTU down either to default of 1500 or lower, say 1200 to allow for any encapsulation that may occur. Would need to be done on MBP too. Largely a stab in the dark but not difficult to change so would be worth a quick try.

Benihana · 13 Oct 2014 at 22:44

Nope all settings on auto now, told extender to get all the settings from the modem. Turned off any security, turned off Jumbo Frames (only available on Synology), checked firewalls were off, lowered MTU on MBP & Modem. No dice.

I think its something with the extender, but it costs as much to replace that as a set of 650mbps powerline adapters. Think I'll give them a try and see if it solves it. Thanks for all your help Cenedd!

Benihana · 23 Oct 2014 at 10:35

Just an update for anyone interested. Went for some Devolo 650+ adapters which are working great. TimeMachine issue solved and internet/wifi has been rock steady ever since. Netgear Extender soon to be on ebay!