Desperate search for Arista EOS firmware

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

IronAddict

New Member
Jul 3, 2021
8
0
1
Hi All,

Am in need of another hint to push me along ...

I've accessed the 7050 switch .. got a putty SSH link into it .. am able to navigate the CLI with some minor commands. Am starting to "remember" linux and the numerous commands and have navigated through the EXEC mode

I have Mellanox X3 NICs (40gps ethernet mode) in both ESXI 5.5 and 7.0 servers -- (so 2 ent interfaces when i do a "show interfaces")
I have also have a "head" with a 10Gbps ethernet card
The ESXI have new vSwitches and the guests now have VMX3 NICs Configured
All green lights on the hardware (yay) AND the vSwitch Network configuration screen

The Arista was added to upgrade a 1gbps network which uses a TP-Link Router Switch (acting as the DHCP and 60 linked smart devices with MAC binding to IP) connected to the internet all configured to 192.168.0.XX subnet -- and a Netgear GS724AT 24 port switch (works well especially after a fan replacement)

All the new VMX3 NICs and i believe by the arista switch (even though i've configured to go to 192.168.0.1 as the network gateway), are getting assigned 169.254.107.XXX addresses -- with all the cascading problems of "unidentified network", "host unreachable" on a ping, etc .. within the 192.168.0.X network

Edit : i did also just RTFM (read the f*****g manual) and realized that the SFP+ ports needed to be activated -- as an FYI to other newbies like me -- the 7050Q-16 can have the SFP+ ports (et17-24) software disabled -- when enabled these 8 SFP+ take the last 2 QSFP+ ports to active themselves ... the command is hardware port-group 1 select Et17-20 (in config mode) i now have green lights on the SFP+ and the transceivers are working on Cat6 cable RJ45 --



HINT **NOT** NEEDED

what/how do i need to do in order to have the Arista disperse 192.168.0.XXX addresses taken from the TP-Link DHCP? IS there a DHCP service that automatically starts in EOS? BTW I've tried to assign fixed 192.168.0.x addresses to the Virtual NICs attached to the Virtual Switch attached to the Arista but that's a no go.

I am reading the manuals but since this is my first foray with this type of technology a lot of the lingo is taken in as a conversation with my dog yields "blah blah blah fido blah blah"

Thanks in advance for the next "push".

The Edit above solved my "problem" -- I.E. once the SFP+ ports were made active the Arista is now connecting to the DHCP and issuing correct IPs .. i was able to ping the VM within one of the ESXI from a networked laptop.

I guess RTFM is the order of the day and this Software switch function much differently than "dumb" managed switches.
 
Last edited:

IronAddict

New Member
Jul 3, 2021
8
0
1
Foray -- yup all Arista "stuff" that i made sure i was following all the other folks on here --

When i do a "show interfaces" both 40gbps ports with cables are showing "up" -- the vSwitch on the ESXIs are showing active

the only one i'm following up now are the SFP+ transceivers that i've put in the Trendnet 10Gbps ports -- although the Trendnet switch i have those plugged in is showing a green light -- the Arista Switch (ethernet17 is showing up as "down") -- cant be that i got 2 separate types of SFP+, both with Arista compatibility that are bad .. ugh

For my question above i'm reading and believe i need to set up a community VLan

**Update** i posted above that my "problem" was the SFP+ ports weren't enabled ... having RTFM figured out what was needed and all good .... for now ... until the next challenge :)
 
Last edited:

IronAddict

New Member
Jul 3, 2021
8
0
1
For those interested i found this little nugget ... still up and running -- i think the 4.29.0.2f is really 4.28... thought i'd post it -- i'm about to check the MD5 to see if its legit ... enjoy while its still there Index of /Arista-EOS/
 

IronAddict

New Member
Jul 3, 2021
8
0
1
yup that's that's what i figured too after expending much google-foo on trying to find the MD5 ... to no avail i might add
 

vangoose

Active Member
May 21, 2019
326
105
43
Canada
I need it too, still running 4.23.
Haven't touched both arista and icx for long time. Once fully configured they just run.
 

Scott Laird

Active Member
Aug 30, 2014
323
156
43
That has been up for a surprisingly long time, at least a few months.
Also, EOS SWI files are just zip files. So you can inspect inside of them fairly easily. That's not to say that SWI files from random places on the internet are safe, but at least you have *some* hope of seeing what's inside of the package.
 

i386

Well-Known Member
Mar 18, 2016
4,411
1,638
113
35
Germany
EOS is Linux, i think IT was Debian based. You can start bash and Run Linux commands...
 

bitbckt

will google compiler errors for scotch
Feb 22, 2022
216
136
43
EOS is Linux, i think IT was Debian based. You can start bash and Run Linux commands...
Here's one sample:

Code:
$ uname -a
Linux malachite 3.4.43.Ar-11723055.41811M #1 SMP PREEMPT Mon Mar 25 15:39:01 PDT 2019 x86_64 x86_64 x86_64 GNU/Linux
$ cat /etc/redhat-release
Fedora release 18 (Spherical Cow)
 

Scott Laird

Active Member
Aug 30, 2014
323
156
43
Also, 4.29.x drops support for another batch of older switches, including the 7050qx-32s, which was the only semi-affordable Arista switch that would run current EOS releases. I wish I'd bought a boatload of 7060s a couple years ago when they were "cheap", before the price went up ~6x.

Oh well. They'll come back down in a couple years. Hopefully.
 

bitbckt

will google compiler errors for scotch
Feb 22, 2022
216
136
43
There’s another year of potential software updates for the 7050QX-32S in the 4.28 train. It’s not quite dead yet.
 

Scott Laird

Active Member
Aug 30, 2014
323
156
43
Yeah, but we shouldn't expect new features past 4.28.2F or so. Which mostly isn't a big deal, but there are some BGP issues that are fixed in 4.29, and python/sshd/etc will start getting old. I have a few old (and even not-that-old) boxes that I can't ssh into because their sshd doesn't support modern ciphers, but that's probably less of an issue with EOS. Upgrading sshd is probably not *that* terrible. I mean, upgrading it is easy; getting it to survive a reboot probably isn't terrible.

In any case, please don't say "run sonic" as an answer. Sonic (SoNIC? SONiC? Can't remember which letter they lowercased this time) seems great, if you're running hundreds of identical devices and designed your network around it. For enterprise-y or home use, it's a giant time sink and just isn't worth it.