Junos Classic vs EVO #1972

jnicholson56 · 2025-01-11T19:51:54Z

jnicholson56
Jan 11, 2025

Is there any guidance on how to handle templates for the two flavors of JUNOS, Classic and EVO? When using the templates there is only juniper_junos as an option. This works fine on both Classic and EVO for the most part. Both versions still run the same copy of Junos (I have verified this with Juniper developers). Most of the command output is the same or has slight differences, but there are some commands that have vastly different output.

The difference between Classic and EVO is the OS underneath Junos.

Classic uses FreeBSD.
EVO uses a customer Linux called Juniper Linux Distribution.

Some commands pull data out of the OS underneath. One such command is show system processes brief. This uses data from the top command in the OS to display CPU and Memory usage info. Due to the differing OSes, the output is pretty unique to each flavor of Junos.

Classic Output:

re0> show system processes brief

last pid: 62713;  load averages:  0.62,  0.47,  0.46  up 36+12:01:55    19:35:23
261 processes: 2 running, 258 sleeping, 1 waiting
CPU:  3.0% user,  0.0% nice,  1.7% system,  0.1% interrupt, 95.2% idle
Mem: 3081M Active, 7342M Inact, 1674M Wired, 578M Buf, 82G Free
Swap: 12G Total, 12G Free

EVO Output (EVO displays all RE's and FPC's at the same time):

re0> show system processes brief
-------------------------------
node: fpc0
-------------------------------
top - 19:42:06 up 193 days, 22 min,  0 users,  load average: 0.63, 1.66, 1.49
Tasks: 273 total,   1 running, 272 sleeping,   0 stopped,   0 zombie
%Cpu(s):  1.9 us,  1.6 sy,  0.0 ni, 96.5 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
MiB Mem :  31554.4 total,  17080.5 free,  10735.4 used,   3738.5 buff/cache
MiB Swap:    100.0 total,    100.0 free,      0.0 used.  19701.3 avail Mem
-------------------------------
node: re0
-------------------------------
top - 19:42:06 up 193 days, 25 min,  1 user,  load average: 1.33, 1.18, 1.13
Tasks: 529 total,   1 running, 528 sleeping,   0 stopped,   0 zombie
%Cpu(s):  4.4 us,  0.9 sy,  0.0 ni, 94.1 id,  0.0 wa,  0.3 hi,  0.3 si,  0.0 st
MiB Mem : 128191.6 total,  65727.8 free,  17833.5 used,  44630.4 buff/cache
MiB Swap:   8192.0 total,   8192.0 free,      0.0 used. 108186.1 avail Mem
-------------------------------
node: re1
-------------------------------
top - 19:42:06 up 193 days, 25 min,  0 users,  load average: 0.27, 0.20, 0.20
Tasks: 401 total,   1 running, 400 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.6 us,  0.6 sy,  0.0 ni, 98.8 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
MiB Mem : 128191.6 total,  91437.6 free,   6202.9 used,  30551.2 buff/cache
MiB Swap:   4096.0 total,   3184.2 free,    911.8 used. 119880.7 avail Mem

As we can see here the output is different enough that someone would naturally gravitate towards making separate templates for each. I do believe it is possible to make one template.

Most of the other Junos based commands are nearly identical across versions. If a new juniper_junos_evo option were added, templates would need to be duplicated etc.

My question here is How do we handle templates for Classic vs EVO?

mjbear · 2025-01-12T13:37:33Z

mjbear
Jan 12, 2025

Is there any guidance on how to handle templates for the two flavors of JUNOS, Classic and EVO?

Since the template chosen via the index file is based on platform and OS type, we would need to handle any differences between Classic and Evolved in the same template.

📝 Note: If this helps and is an answer, please mark it as such. Thank you!

When using the templates there is only juniper_junos as an option.

For now I'm only aware of the single NOS being "junos".

This works fine on both Classic and EVO for the most part. Both versions still run the same copy of Junos (I have verified this with Juniper developers). Most of the command output is the same or has slight differences, but there are some commands that have vastly different output.

This is the ideal part in that since it normally "works fine on both Classic and EVO", we (community) can do our best to make the template(s) support both.

The difference between Classic and EVO is the OS underneath Junos.
* Classic uses FreeBSD.

* EVO uses a customer Linux called `Juniper Linux Distribution`.

👍 Agree.

Some commands pull data out of the OS underneath. One such command is show system processes brief. This uses data from the top command in the OS to display CPU and Memory usage info. Due to the differing OSes, the output is pretty unique to each flavor of Junos.

I don't have access to any JunOS images presently (I need to go look at jLabs and/or talk to Juniper), but the JunOS display filter might also come in handy if certain templates get to be a pain. The gotcha is/was that not every command has this display support.

https://www.juniper.net/documentation/us/en/software/junos/cli/topics/topic-map/filtering-operational-command.html#id-pipe-filter-functions-in-the-junos-os-command-line-interface__d11e371

... snipped ...

As we can see here the output is different enough that someone would naturally gravitate towards making separate templates for each. I do believe it is possible to make one template.

Most of the other Junos based commands are nearly identical across versions. If a new juniper_junos_evo option were added, templates would need to be duplicated etc.

My question here is How do we handle templates for Classic vs EVO?

To somewhat reiterate from above, there's a single NOS juniper_junos so we'll use what they have in common and make the rest work.

Where we can it would likely be smart to separate the EVO-specific rules.

# simplified example, haha
Value X (\S+)

Start
  ^some\s+${X}\s+match\s*$$
  # Evolved specific
  ^some\s+other\s+Evo\s+specific\s+${X}\s+match\s*$$

0 replies

jmcgill298 · 2025-01-13T19:02:29Z

jmcgill298
Jan 13, 2025
Maintainer

If there needs to be a new OS added, that can be done. We prefer to align with what works with Netmiko though. The use of States is a good way to handle widely different output, such as Cisco IOS show mac address command (look at the template)

2 replies

mjbear Jan 13, 2025

The use of States is a good way to handle widely different output, such as Cisco IOS show mac address command (look at the template)

💯%
I hadn't update my response above, but in a direct message I mentioned to Joseph that States would be another possibility.
Glad we agree, hehe.

jnicholson56 Jan 13, 2025
Author

If there needs to be a new OS added, that can be done. We prefer to align with what works with Netmiko though. The use of States is a good way to handle widely different output, such as Cisco IOS show mac address command (look at the template)

Oh I see. Textfsm is more advanced than I realized. I have not seen states before. Got it.

jnicholson56 · 2025-01-13T19:36:22Z

jnicholson56
Jan 13, 2025
Author

Oh thanks for the link. I was looking at other Junos templates to see if something had been done to handle multiple OS versions. Adding a new OS can be a bit burdensome. Most of the juniper_junos templates overlap with no issues and having to replicate them across two versions would be confusing and a pain I think. Michael also helped me understand that templates can take comments too. I can make some notes within the template to make it clear that there appear to be two sets of code for two versions. I will mock something up and paste it here when I get a chance in the next few days to get some feedback on the direction I am taking. Thanks!

…

On Mon, Jan 13, 2025 at 1:07 PM Michael Bear ***@***.***> wrote: The use of States is a good way to handle widely different output, such as Cisco IOS show mac address command (look at the template) 💯% I hadn't update my response above, but in a direct message I mentioned to Joseph that States would be another possibility. Glad we agree, hehe. — Reply to this email directly, view it on GitHub <#1972 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AILVUEXRZ7DULHI6SOV6N432KQFGZAVCNFSM6AAAAABVAIJKFCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBSGM3TMMA> . You are receiving this because you authored the thread.Message ID: <networktocode/ntc-templates/repo-discussions/1972/comments/11823760@ github.com>

0 replies

jnicholson56 · 2025-01-13T19:41:24Z

jnicholson56
Jan 13, 2025
Author

By "link" I meant a suggestion to the Cisco template. Sorry that read weird when I looked over it again. :D On Mon, Jan 13, 2025 at 1:36 PM Joseph Nicholson ***@***.***> wrote:

…

Oh thanks for the link. I was looking at other Junos templates to see if something had been done to handle multiple OS versions. Adding a new OS can be a bit burdensome. Most of the juniper_junos templates overlap with no issues and having to replicate them across two versions would be confusing and a pain I think. Michael also helped me understand that templates can take comments too. I can make some notes within the template to make it clear that there appear to be two sets of code for two versions. I will mock something up and paste it here when I get a chance in the next few days to get some feedback on the direction I am taking. Thanks! On Mon, Jan 13, 2025 at 1:07 PM Michael Bear ***@***.***> wrote: > The use of States is a good way to handle widely different output, such > as Cisco IOS show mac address command (look at the template) > > 💯% > I hadn't update my response above, but in a direct message I mentioned to > Joseph that States would be another possibility. > Glad we agree, hehe. > > — > Reply to this email directly, view it on GitHub > <#1972 (reply in thread)>, > or unsubscribe > <https://github.com/notifications/unsubscribe-auth/AILVUEXRZ7DULHI6SOV6N432KQFGZAVCNFSM6AAAAABVAIJKFCVHI2DSMVQWIX3LMV43URDJONRXK43TNFXW4Q3PNVWWK3TUHMYTCOBSGM3TMMA> > . > You are receiving this because you authored the thread.Message ID: > <networktocode/ntc-templates/repo-discussions/1972/comments/11823760@ > github.com> >

0 replies

jnicholson56 · 2025-01-13T20:06:30Z

jnicholson56
Jan 13, 2025
Author

I quickly put this together using the nornir site. When I feed it actual router output for EVO or Classic is parses as expected. I see how to make multiple test files as well to make sure each gets tested properly. Thanks guys I think you got my questions answered.

Value NODE (\S+)
Value MEMORY_FORMAT (\w)
Value SWAP_FORMAT (\w)
Value MEMORY_ACTIVE (\S+)
Value MEMORY_FREE (\S+)
Value MEMORY_USED (\S+)
Value MEMORY_BUFFER (\S+)
Value SWAP_TOTAL (\S+)
Value SWAP_FREE (\S+)
Value SWAP_USED (\S+)
Value SWAP_AVAIL (\S+)

Start
  ^-+\s*$$ -> TYPE1
  ^last\s*pid: -> TYPE2

# JUNOS EVO template
TYPE1
  ^-+\s*$$
  ^node:\s*${NODE}
  ^top\s*
  ^Tasks:\s*
  ^\%Cpu
  ^${MEMORY_FORMAT}iB\s+(Mem|MEMORY)\s+:\s+${MEMORY_ACTIVE}\s+total,\s+${MEMORY_FREE}\s+free,\s+${MEMORY_USED}\s+used,\s+${MEMORY_BUFFER}\s+buff/cache\s*
  ^${SWAP_FORMAT}iB\s+Swap:\s+${SWAP_TOTAL}\s+total,\s+${SWAP_FREE}\s+free,\s+${SWAP_USED}\s+used\.\s+${SWAP_AVAIL}\s+avail\s+(Mem|MEMORY)\s* -> Record
  ^\s*$$
  ^. -> Error

# Junos Classic template
TYPE2
  ^last pid:
  ^\d+\s+processes:\s+\d+\s+running,\s+\d+\s+sleeping,\s+\d+\s+zombie,\s+\d+\s+waiting\s*$$
  ^CPU:\s+\d+\.\d+\%\s+user,\s+\d+\.\d+\%\s+nice,\s+\d+\.\d+\%\s+system,\s+\d+\.\d+\%\s+interrupt,\s+\d+\.\d+\%\s+idle\s*$$
  ^Mem:\s+${MEMORY_ACTIVE}\s+Active,\s+${MEMORY_FREE}\s+Inact,\s+${MEMORY_USED}\s+Wired,\s+${MEMORY_BUFFER}\s+Buf,\s+${SWAP_AVAIL}\s+Free\s*$$
  ^Swap:\s+${SWAP_TOTAL}\s+Total,\s+${SWAP_FREE}\s+Free\s*$$
  ^. -> Error

0 replies

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Junos Classic vs EVO #1972

{{title}}

Replies: 5 comments 2 replies

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

{{title}}

Select a reply

Junos Classic vs EVO #1972

jnicholson56 Jan 11, 2025

Replies: 5 comments · 2 replies

mjbear Jan 12, 2025

jmcgill298 Jan 13, 2025 Maintainer

mjbear Jan 13, 2025

jnicholson56 Jan 13, 2025 Author

jnicholson56 Jan 13, 2025 Author

jnicholson56 Jan 13, 2025 Author

jnicholson56 Jan 13, 2025 Author

jnicholson56
Jan 11, 2025

Replies: 5 comments 2 replies

mjbear
Jan 12, 2025

jmcgill298
Jan 13, 2025
Maintainer

jnicholson56 Jan 13, 2025
Author

jnicholson56
Jan 13, 2025
Author

jnicholson56
Jan 13, 2025
Author

jnicholson56
Jan 13, 2025
Author