Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
change
sglang
Commits
8b33d8df
Unverified
Commit
8b33d8df
authored
May 26, 2025
by
wangxiyu191
Committed by
GitHub
May 26, 2025
Browse files
[PD] Fix prefill_servers in mini_lb (#6527)
parent
e235be16
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
10 additions
and
3 deletions
+10
-3
python/sglang/srt/disaggregation/mini_lb.py
python/sglang/srt/disaggregation/mini_lb.py
+10
-3
No files found.
python/sglang/srt/disaggregation/mini_lb.py
View file @
8b33d8df
...
@@ -50,6 +50,13 @@ class MiniLoadBalancer:
...
@@ -50,6 +50,13 @@ class MiniLoadBalancer:
self
.
prefill_servers
=
[
p
.
url
for
p
in
prefill_configs
]
self
.
prefill_servers
=
[
p
.
url
for
p
in
prefill_configs
]
self
.
decode_servers
=
decode_servers
self
.
decode_servers
=
decode_servers
def
add_prefill_server
(
self
,
new_prefill_config
:
PrefillConfig
):
self
.
prefill_configs
.
append
(
new_prefill_config
)
self
.
prefill_servers
.
append
(
new_prefill_config
.
url
)
def
add_decode_server
(
self
,
new_decode_server
:
str
):
self
.
decode_servers
.
append
(
new_decode_server
)
def
select_pair
(
self
):
def
select_pair
(
self
):
# TODO: return some message instead of panic
# TODO: return some message instead of panic
assert
len
(
self
.
prefill_configs
)
>
0
,
"No prefill servers available"
assert
len
(
self
.
prefill_configs
)
>
0
,
"No prefill servers available"
...
@@ -157,7 +164,7 @@ class MiniLoadBalancer:
...
@@ -157,7 +164,7 @@ class MiniLoadBalancer:
app
=
FastAPI
()
app
=
FastAPI
()
load_balancer
=
None
load_balancer
:
Optional
[
MiniLoadBalancer
]
=
None
@
app
.
get
(
"/health"
)
@
app
.
get
(
"/health"
)
...
@@ -331,14 +338,14 @@ async def get_models():
...
@@ -331,14 +338,14 @@ async def get_models():
@
app
.
post
(
"/register"
)
@
app
.
post
(
"/register"
)
async
def
register
(
obj
:
PDRegistryRequest
):
async
def
register
(
obj
:
PDRegistryRequest
):
if
obj
.
mode
==
"prefill"
:
if
obj
.
mode
==
"prefill"
:
load_balancer
.
prefill_
configs
.
append
(
load_balancer
.
add_
prefill_
server
(
PrefillConfig
(
obj
.
registry_url
,
obj
.
bootstrap_port
)
PrefillConfig
(
obj
.
registry_url
,
obj
.
bootstrap_port
)
)
)
logger
.
info
(
logger
.
info
(
f
"Registered prefill server:
{
obj
.
registry_url
}
with bootstrap port:
{
obj
.
bootstrap_port
}
"
f
"Registered prefill server:
{
obj
.
registry_url
}
with bootstrap port:
{
obj
.
bootstrap_port
}
"
)
)
elif
obj
.
mode
==
"decode"
:
elif
obj
.
mode
==
"decode"
:
load_balancer
.
decode_server
s
.
append
(
obj
.
registry_url
)
load_balancer
.
add_
decode_server
(
obj
.
registry_url
)
logger
.
info
(
f
"Registered decode server:
{
obj
.
registry_url
}
"
)
logger
.
info
(
f
"Registered decode server:
{
obj
.
registry_url
}
"
)
else
:
else
:
raise
HTTPException
(
raise
HTTPException
(
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment