Skip to content

DTrace and log polish #1773

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 4 commits into from
Sep 29, 2025
Merged

DTrace and log polish #1773

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
b6ca6d8
DTrace polish, move some log messages to warn
Sep 24, 2025
094fca8
missed one
Sep 24, 2025
81612f7
Bump the 1k to 2k
Sep 24, 2025
6efcdb2
final polish
Sep 25, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
65 changes: 22 additions & 43 deletions cmon/src/main.rs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,9 +8,7 @@ use strum::IntoEnumIterator;
use strum_macros::EnumIter;
use tokio::time::{sleep, Duration};

use crucible::{
ClientStopReason, ConnectionMode, DsState, DtraceInfo, NegotiationState,
};
use crucible::DtraceInfo;

/// Connect to crucible control server
#[derive(Parser, Debug)]
Expand Down Expand Up @@ -86,44 +84,25 @@ enum Action {
Repair,
}

/// Translate a [`DsState`] into a three letter string for printing.
fn short_state(dss: DsState) -> String {
/// Translate what the default DsState string is (that we are getting from DTrace)
/// into a three letter string for printing.
fn short_state(dss: &str) -> String {
match dss {
DsState::Connecting {
state: NegotiationState::WaitQuorum,
..
} => "WQ".to_string(),
DsState::Connecting {
state: NegotiationState::Reconcile,
..
} => "REC".to_string(),
DsState::Active => "ACT".to_string(),
DsState::Connecting {
state: NegotiationState::LiveRepairReady,
..
} => "LRR".to_string(),
DsState::Stopping(ClientStopReason::NegotiationFailed(..))
| DsState::Connecting {
mode: ConnectionMode::New,
..
} => "NEW".to_string(),
DsState::Connecting {
mode: ConnectionMode::Faulted,
..
}
| DsState::Stopping(ClientStopReason::Fault(..)) => "FLT".to_string(),
DsState::LiveRepair => "LR".to_string(),
DsState::Connecting {
mode: ConnectionMode::Offline,
..
} => "OFL".to_string(),
DsState::Stopping(ClientStopReason::Deactivated) => "DAV".to_string(),
DsState::Stopping(ClientStopReason::Disabled) => "DIS".to_string(),
DsState::Stopping(ClientStopReason::Replacing)
| DsState::Connecting {
mode: ConnectionMode::Replaced,
..
} => "RPL".to_string(),
"Active" => "ACT".to_string(),
"WaitQuorum" => "WQ".to_string(),
"Reconcile" => "REC".to_string(),
"LiveRepairReady" => "LRR".to_string(),
"New" => "NEW".to_string(),
"Faulted" => "FLT".to_string(),
"Offline" => "OFL".to_string(),
"Replaced" => "RPL".to_string(),
"LiveRepair" => "LR".to_string(),
"Replacing" => "RPC".to_string(),
"Disabled" => "DIS".to_string(),
"Deactivated" => "DAV".to_string(),
"NegotiationFailed" => "NF".to_string(),
"Fault" => "FLT".to_string(),
x => x.to_string(),
}
}

Expand Down Expand Up @@ -277,9 +256,9 @@ fn print_dtrace_row(
DtraceDisplay::State => {
print!(
" {:>3} {:>3} {:>3}",
short_state(d_out.ds_state[0]),
short_state(d_out.ds_state[1]),
short_state(d_out.ds_state[2]),
short_state(&d_out.ds_state[0]),
short_state(&d_out.ds_state[1]),
short_state(&d_out.ds_state[2]),
);
}
DtraceDisplay::UpCount => {
Expand Down
4 changes: 2 additions & 2 deletions tools/dtrace/all_downstairs.d
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -57,9 +57,9 @@ crucible_downstairs*:::work-done

tick-4s
{
printf("%5s %4s %4s %4s %4s %5s %5s %5s %5s %5s\n",
printf("%5s %5s %5s %5s %5s %5s %5s %5s %5s %5s\n",
"PID", "F>", "F<", "W>", "W<", "R>", "R<", "WS", "WIP", "WD");
printa("%05d %@4u %@4u %@4u %@4u %@5u %@5u %@5u %@5u %@5u\n",
printa("%05d %@5u %@5u %@5u %@5u %@5u %@5u %@5u %@5u %@5u\n",
@sf_start, @sf_done, @sw_start, @sw_done, @sr_start, @sr_done,
@work_start, @work_process, @work_done
);
Expand Down
33 changes: 18 additions & 15 deletions tools/dtrace/get-ds-state.d
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,37 +3,40 @@
* Exit after 5 seconds.
*/
#pragma D option quiet
#pragma D option strsize=1k
#pragma D option strsize=2k

/*
* Translate the longer state string into a shorter version
*/
inline string short_state[string ss] =
ss == "active" ? "ACT" :
ss == "new" ? "NEW" :
ss == "replaced" ? "RPL" :
ss == "live_repair_ready" ? "LRR" :
ss == "live_repair" ? "LR" :
ss == "faulted" ? "FLT" :
ss == "offline" ? "OFL" :
ss == "reconcile" ? "REC" :
ss == "wait_quorum" ? "WQ" :
ss == "wait_active" ? "WA" :
ss == "connecting" ? "CON" :
ss == "Active" ? "ACT" :
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there a way to share this function with all the other scripts?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe... Let me see if I can find an easy way.
I think it's "possible", but I don't want the callers of the scripts have to do too much, or remember too much.

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let me see if I can find an easier way.. in a follow up commit :)

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I want these changes so I can triage what is going on when multiple downstairs come and go while doing LR

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You could look at what OPTE does, which has a few shared type and function definitions for these kinds of things.

ss == "WaitQuorum" ? "WQ" :
ss == "Reconcile" ? "REC" :
ss == "LiveRepairReady" ? "LRR" :
ss == "New" ? "NEW" :
ss == "Faulted" ? "FLT" :
ss == "Offline" ? "OFL" :
ss == "LiveRepair" ? "LR" :
ss == "Replacing" ? "RPC" :
ss == "Replaced" ? "RPL" :
ss == "Disabled" ? "DIS" :
ss == "Deactivated" ? "DAV" :
ss == "NegotiationFailed" ? "NF" :
ss == "Fault" ? "FLT" :
ss;

crucible_upstairs*:::up-status
{
my_id = json(copyinstr(arg1), "ok.upstairs_id");
my_sesh = json(copyinstr(arg1), "ok.session_id");

this->ds0state = json(copyinstr(arg1), "ok.ds_state[0].type");
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0]");
this->d0 = short_state[this->ds0state];

this->ds1state = json(copyinstr(arg1), "ok.ds_state[1].type");
this->ds1state = json(copyinstr(arg1), "ok.ds_state[1]");
this->d1 = short_state[this->ds1state];

this->ds2state = json(copyinstr(arg1), "ok.ds_state[2].type");
this->ds2state = json(copyinstr(arg1), "ok.ds_state[2]");
this->d2 = short_state[this->ds2state];

printf("%6d %8s %8s %3s %3s %3s\n",
Expand Down
33 changes: 18 additions & 15 deletions tools/dtrace/get-up-state.d
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Display Upstairs status for all matching processes
*/
#pragma D option quiet
#pragma D option strsize=1k
#pragma D option strsize=2k

/*
* Print the header right away
Expand Down Expand Up @@ -34,17 +34,20 @@ tick-10s
* Translate the longer state string into a shorter version
*/
inline string short_state[string ss] =
ss == "active" ? "ACT" :
ss == "new" ? "NEW" :
ss == "replaced" ? "RPL" :
ss == "live_repair_ready" ? "LRR" :
ss == "live_repair" ? "LR" :
ss == "faulted" ? "FLT" :
ss == "offline" ? "OFL" :
ss == "reconcile" ? "REC" :
ss == "wait_quorum" ? "WQ" :
ss == "wait_active" ? "WA" :
ss == "connecting" ? "CON" :
ss == "Active" ? "ACT" :
ss == "WaitQuorum" ? "WQ" :
ss == "Reconcile" ? "REC" :
ss == "LiveRepairReady" ? "LRR" :
ss == "New" ? "NEW" :
ss == "Faulted" ? "FLT" :
ss == "Offline" ? "OFL" :
ss == "LiveRepair" ? "LR" :
ss == "Replacing" ? "RPC" :
ss == "Replaced" ? "RPL" :
ss == "Disabled" ? "DIS" :
ss == "Deactivated" ? "DAV" :
ss == "NegotiationFailed" ? "NF" :
ss == "Fault" ? "FLT" :
ss;

/*
Expand All @@ -54,13 +57,13 @@ inline string short_state[string ss] =
*/
crucible_upstairs*:::up-status
{
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0].type");
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0]");
this->d0 = short_state[this->ds0state];

this->ds1state = json(copyinstr(arg1), "ok.ds_state[1].type");
this->ds1state = json(copyinstr(arg1), "ok.ds_state[1]");
this->d1 = short_state[this->ds1state];

this->ds2state = json(copyinstr(arg1), "ok.ds_state[2].type");
this->ds2state = json(copyinstr(arg1), "ok.ds_state[2]");
this->d2 = short_state[this->ds2state];

this->full_session_id = json(copyinstr(arg1), "ok.session_id");
Expand Down
33 changes: 18 additions & 15 deletions tools/dtrace/simple.d
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
#pragma D option quiet
#pragma D option strsize=1k
#pragma D option strsize=2k

dtrace:::BEGIN
{
Expand Down Expand Up @@ -28,17 +28,20 @@ dtrace:::BEGIN, tick-20s
* Translate the longer state string into a shorter version
*/
inline string short_state[string ss] =
ss == "active" ? "ACT" :
ss == "new" ? "NEW" :
ss == "live_repair_ready" ? "LRR" :
ss == "live_repair" ? "LR" :
ss == "faulted" ? "FLT" :
ss == "offline" ? "OFL" :
ss == "reconcile" ? "REC" :
ss == "wait_quorum" ? "WQ" :
ss == "wait_active" ? "WA" :
ss == "replaced" ? "RPL" :
ss == "connecting" ? "CON" :
ss == "Active" ? "ACT" :
ss == "WaitQuorum" ? "WQ" :
ss == "Reconcile" ? "REC" :
ss == "LiveRepairReady" ? "LRR" :
ss == "New" ? "NEW" :
ss == "Faulted" ? "FLT" :
ss == "Offline" ? "OFL" :
ss == "LiveRepair" ? "LR" :
ss == "Replacing" ? "RPC" :
ss == "Replaced" ? "RPL" :
ss == "Disabled" ? "DIS" :
ss == "Deactivated" ? "DAV" :
ss == "NegotiationFailed" ? "NF" :
ss == "Fault" ? "FLT" :
ss;

/*
Expand All @@ -48,13 +51,13 @@ inline string short_state[string ss] =
*/
crucible_upstairs*:::up-status
{
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0].type");
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0]");
this->d0 = short_state[this->ds0state];

this->ds1state = json(copyinstr(arg1), "ok.ds_state[1].type");
this->ds1state = json(copyinstr(arg1), "ok.ds_state[1]");
this->d1 = short_state[this->ds1state];

this->ds2state = json(copyinstr(arg1), "ok.ds_state[2].type");
this->ds2state = json(copyinstr(arg1), "ok.ds_state[2]");
this->d2 = short_state[this->ds2state];

this->full_upstairs_id = json(copyinstr(arg1), "ok.upstairs_id");
Expand Down
33 changes: 18 additions & 15 deletions tools/dtrace/single_up_info.d
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@
* Display internal Upstairs status for the PID provided as $1
*/
#pragma D option quiet
#pragma D option strsize=1k
#pragma D option strsize=2k
/*
* Print the header right away
*/
Expand Down Expand Up @@ -40,17 +40,20 @@ dtrace:::BEGIN, tick-1s
* Translate the longer state string into a shorter version
*/
inline string short_state[string ss] =
ss == "active" ? "ACT" :
ss == "new" ? "NEW" :
ss == "live_repair_ready" ? "LRR" :
ss == "live_repair" ? "LR" :
ss == "faulted" ? "FLT" :
ss == "offline" ? "OFL" :
ss == "reconcile" ? "REC" :
ss == "wait_quorum" ? "WQ" :
ss == "wait_active" ? "WA" :
ss == "replaced" ? "RPL" :
ss == "connecting" ? "CON" :
ss == "Active" ? "ACT" :
ss == "WaitQuorum" ? "WQ" :
ss == "Reconcile" ? "REC" :
ss == "LiveRepairReady" ? "LRR" :
ss == "New" ? "NEW" :
ss == "Faulted" ? "FLT" :
ss == "Offline" ? "OFL" :
ss == "LiveRepair" ? "LR" :
ss == "Replacing" ? "RPC" :
ss == "Replaced" ? "RPL" :
ss == "Disabled" ? "DIS" :
ss == "Deactivated" ? "DAV" :
ss == "NegotiationFailed" ? "NF" :
ss == "Fault" ? "FLT" :
ss;

crucible_upstairs*:::up-status
Expand All @@ -61,13 +64,13 @@ crucible_upstairs*:::up-status
* All these local variables require the "this->" so the probe firing
* from different sessions don't collide with each other.
*/
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0].type");
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0]");
this->d0 = short_state[this->ds0state];

this->ds1state = json(copyinstr(arg1), "ok.ds_state[1].type");
this->ds1state = json(copyinstr(arg1), "ok.ds_state[1]");
this->d1 = short_state[this->ds1state];

this->ds2state = json(copyinstr(arg1), "ok.ds_state[2].type");
this->ds2state = json(copyinstr(arg1), "ok.ds_state[2]");
this->d2 = short_state[this->ds2state];

this->full_upstairs_id = json(copyinstr(arg1), "ok.upstairs_id");
Expand Down
33 changes: 18 additions & 15 deletions tools/dtrace/sled_upstairs_info.d
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,7 @@
* will share the PID, but have unique SESSIONs.
*/
#pragma D option quiet
#pragma D option strsize=1k
#pragma D option strsize=2k
/*
* Print the header right away
*/
Expand Down Expand Up @@ -40,17 +40,20 @@ tick-1s
* Translate the longer state string into a shorter version
*/
inline string short_state[string ss] =
ss == "active" ? "ACT" :
ss == "new" ? "NEW" :
ss == "live_repair_ready" ? "LRR" :
ss == "live_repair" ? "LR" :
ss == "faulted" ? "FLT" :
ss == "offline" ? "OFL" :
ss == "reconcile" ? "REC" :
ss == "wait_quorum" ? "WQ" :
ss == "wait_active" ? "WA" :
ss == "replaced" ? "RPL" :
ss == "connecting" ? "CON" :
ss == "Active" ? "ACT" :
ss == "WaitQuorum" ? "WQ" :
ss == "Reconcile" ? "REC" :
ss == "LiveRepairReady" ? "LRR" :
ss == "New" ? "NEW" :
ss == "Faulted" ? "FLT" :
ss == "Offline" ? "OFL" :
ss == "LiveRepair" ? "LR" :
ss == "Replacing" ? "RPC" :
ss == "Replaced" ? "RPL" :
ss == "Disabled" ? "DIS" :
ss == "Deactivated" ? "DAV" :
ss == "NegotiationFailed" ? "NF" :
ss == "Fault" ? "FLT" :
ss;

crucible_upstairs*:::up-status
Expand All @@ -59,13 +62,13 @@ crucible_upstairs*:::up-status
this->upstairs_id = json(copyinstr(arg1), "ok.upstairs_id");
this->session_id = json(copyinstr(arg1), "ok.session_id");

this->ds0state = json(copyinstr(arg1), "ok.ds_state[0].type");
this->ds0state = json(copyinstr(arg1), "ok.ds_state[0]");
this->d0 = short_state[this->ds0state];

this->ds1state = json(copyinstr(arg1), "ok.ds_state[1].type");
this->ds1state = json(copyinstr(arg1), "ok.ds_state[1]");
this->d1 = short_state[this->ds1state];

this->ds2state = json(copyinstr(arg1), "ok.ds_state[2].type");
this->ds2state = json(copyinstr(arg1), "ok.ds_state[2]");
this->d2 = short_state[this->ds2state];


Expand Down
Loading