fix: update clickhouse queries

Signed-off-by: Shivanshu Raj Shrivastava <shivanshu1333@gmail.com>
This commit is contained in:
Shivanshu Raj Shrivastava 2025-06-01 22:38:56 +05:30
parent 40ca9adbfc
commit 3aacb99adc
No known key found for this signature in database
GPG Key ID: D34D26C62AC3E9AE

View File

@ -42,11 +42,8 @@ FROM (
(serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[10]s)
)
GROUP BY trace_id
) AS funnel
WHERE
t2_time > t1_time
AND t1_time > 0
AND t2_time > 0
HAVING t1_time > 0 AND t2_time > t1_time
)
ORDER BY t1_time
LIMIT 5;`
return fmt.Sprintf(queryTemplate,
@ -104,17 +101,12 @@ FROM (
timestamp BETWEEN start_ts AND end_ts
AND (
(serviceName = step1.1 AND name = step1.2 AND (contains_error_t1 = 0 OR has_error = true) %[12]s)
OR
(serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[13]s)
OR
(serviceName = step3.1 AND name = step3.2 AND (contains_error_t3 = 0 OR has_error = true) %[14]s)
OR (serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[13]s)
OR (serviceName = step3.1 AND name = step3.2 AND (contains_error_t3 = 0 OR has_error = true) %[14]s)
)
GROUP BY trace_id
) AS funnel
WHERE
t1_time > 0
AND t2_time > t1_time
AND t3_time > t2_time
HAVING t1_time > 0 AND t2_time > t1_time AND t3_time > t2_time
)
ORDER BY t1_time
LIMIT 5;`
return fmt.Sprintf(queryTemplate,
@ -167,8 +159,8 @@ WITH
trace_id,
minIf(timestamp, serviceName = step1.1 AND name = step1.2) AS t1_time,
minIf(timestamp, serviceName = step2.1 AND name = step2.2) AS t2_time,
anyIf(has_error, serviceName=step1.1 AND name=step1.2) AS s1_error,
anyIf(has_error, serviceName=step2.1 AND name=step2.2) AS s2_error
toUInt8(anyIf(has_error, serviceName = step1.1 AND name = step1.2)) AS s1_error,
toUInt8(anyIf(has_error, serviceName = step2.1 AND name = step2.2)) AS s2_error
FROM signoz_traces.signoz_index_v3
WHERE
timestamp BETWEEN start_ts AND end_ts
@ -178,13 +170,13 @@ WITH
(serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[12]s)
)
GROUP BY trace_id
HAVING t1_time > 0 AND t2_time > t1_time
)
, totals AS (
SELECT
countIf(t1_time>0) AS total_s1_spans,
countIf(t1_time>0 AND t2_time>t1_time) AS total_s2_spans,
countIf(t1_time>0 AND t2_time>0) AS total_s2_raw_spans,
count() AS total_s1_spans,
countIf(t2_time > t1_time) AS total_s2_spans,
sum(s1_error) AS sum_s1_error,
sum(s2_error) AS sum_s2_error,
avg((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS avg_duration,
@ -193,8 +185,8 @@ WITH
)
SELECT
round(total_s2_spans * 100.0 / total_s1_spans, 2) AS conversion_rate,
total_s2_raw_spans / time_window_sec AS avg_rate,
round(if(total_s1_spans > 0, total_s2_spans * 100.0 / total_s1_spans, 0), 2) AS conversion_rate,
total_s2_spans / time_window_sec AS avg_rate,
greatest(sum_s1_error, sum_s2_error) AS errors,
avg_duration,
latency
@ -257,9 +249,9 @@ WITH
minIf(timestamp, serviceName = step1.1 AND name = step1.2) AS t1_time,
minIf(timestamp, serviceName = step2.1 AND name = step2.2) AS t2_time,
minIf(timestamp, serviceName = step3.1 AND name = step3.2) AS t3_time,
anyIf(has_error, serviceName=step1.1 AND name=step1.2) AS s1_error,
anyIf(has_error, serviceName=step2.1 AND name=step2.2) AS s2_error,
anyIf(has_error, serviceName=step3.1 AND name=step3.2) AS s3_error
toUInt8(anyIf(has_error, serviceName = step1.1 AND name = step1.2)) AS s1_error,
toUInt8(anyIf(has_error, serviceName = step2.1 AND name = step2.2)) AS s2_error,
toUInt8(anyIf(has_error, serviceName = step3.1 AND name = step3.2)) AS s3_error
FROM signoz_traces.signoz_index_v3
WHERE
timestamp BETWEEN start_ts AND end_ts
@ -275,24 +267,30 @@ WITH
SELECT
countIf(t1_time > 0) AS total_s1_spans,
countIf(t1_time > 0 AND t2_time > t1_time) AS total_s2_spans,
countIf(t1_time > 0 AND t2_time > t1_time AND t3_time > t2_time) AS total_s3_spans,
countIf(t2_time > 0 AND t3_time > t2_time) AS total_s3_spans,
sum(s1_error) AS sum_s1_error,
sum(s2_error) AS sum_s2_error,
sum(s3_error) AS sum_s3_error,
avg((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS avg_duration,
quantile(0.99)((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS latency
avgIf((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time))/1e6, t1_time > 0 AND t2_time > t1_time) AS avg_duration_12,
quantileIf(0.99)((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time))/1e6, t1_time > 0 AND t2_time > t1_time) AS latency_12,
avgIf((toUnixTimestamp64Nano(t3_time) - toUnixTimestamp64Nano(t2_time))/1e6, t2_time > 0 AND t3_time > t2_time) AS avg_duration_23,
quantileIf(0.99)((toUnixTimestamp64Nano(t3_time) - toUnixTimestamp64Nano(t2_time))/1e6, t2_time > 0 AND t3_time > t2_time) AS latency_23
FROM funnel
)
SELECT
round(total_s3_spans * 100.0 / total_s1_spans, 2) AS conversion_rate,
total_s2_spans / time_window_sec AS avg_rate,
round(if(total_s1_spans > 0, total_s3_spans * 100.0 / total_s1_spans, 0), 2) AS conversion_rate,
total_s3_spans / nullIf(time_window_sec, 0) AS avg_rate,
greatest(sum_s1_error, sum_s2_error, sum_s3_error) AS errors,
avg_duration,
latency
avg_duration_23 AS avg_duration,
latency_23 AS latency
FROM totals;
`
return fmt.Sprintf(queryTemplate,
return fmt.Sprintf(
queryTemplate,
containsErrorT1,
containsErrorT2,
containsErrorT3,
@ -336,10 +334,10 @@ WITH
('%[7]s','%[8]s') AS step2
SELECT
countIf(t1_time>0) AS total_s1_spans,
count() AS total_s1_spans,
countIf(t1_error = 1) AS total_s1_errored_spans,
countIf(t1_time>0 AND t2_time>t1_time) AS total_s2_spans,
countIf(t1_time>0 AND t2_time>t1_time AND t2_error=1) AS total_s2_errored_spans
countIf(t2_time > t1_time) AS total_s2_spans,
countIf(t2_time > t1_time AND t2_error = 1) AS total_s2_errored_spans
FROM (
SELECT
trace_id,
@ -356,8 +354,9 @@ FROM (
(serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[10]s)
)
GROUP BY trace_id
) AS funnel
WHERE t1_time>0;`
HAVING t1_time > 0 AND t2_time > t1_time
) AS funnel;
`
return fmt.Sprintf(queryTemplate,
containsErrorT1,
containsErrorT2,
@ -401,12 +400,12 @@ WITH
('%[10]s','%[11]s') AS step3
SELECT
countIf(t1_time>0) AS total_s1_spans,
count() AS total_s1_spans,
countIf(t1_error = 1) AS total_s1_errored_spans,
countIf(t1_time>0 AND t2_time>t1_time) AS total_s2_spans,
countIf(t1_time>0 AND t2_time>t1_time AND t2_error=1) AS total_s2_errored_spans,
countIf(t1_time>0 AND t2_time>t1_time AND t3_time>t2_time) AS total_s3_spans,
countIf(t1_time>0 AND t2_time>t1_time AND t3_time>t2_time AND t3_error=1) AS total_s3_errored_spans
countIf(t2_time > t1_time) AS total_s2_spans,
countIf(t2_time > t1_time AND t2_error = 1) AS total_s2_errored_spans,
countIf(t2_time > t1_time AND t3_time > t2_time) AS total_s3_spans,
countIf(t2_time > t1_time AND t3_time > t2_time AND t3_error = 1) AS total_s3_errored_spans
FROM (
SELECT
trace_id,
@ -421,14 +420,13 @@ FROM (
timestamp BETWEEN start_ts AND end_ts
AND (
(serviceName = step1.1 AND name = step1.2 AND (contains_error_t1 = 0 OR has_error = true) %[12]s)
OR
(serviceName=step2.1 AND name=step2.2 AND (contains_error_t2=0 OR has_error=true) %[13]s)
OR
(serviceName=step3.1 AND name=step3.2 AND (contains_error_t3=0 OR has_error=true) %[14]s)
OR (serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[13]s)
OR (serviceName = step3.1 AND name = step3.2 AND (contains_error_t3 = 0 OR has_error = true) %[14]s)
)
GROUP BY trace_id
) AS funnel
WHERE t1_time>0;`
HAVING t1_time > 0 AND t2_time > t1_time AND t3_time > t2_time
) AS funnel;
`
return fmt.Sprintf(queryTemplate,
containsErrorT1,
containsErrorT2,
@ -488,10 +486,11 @@ FROM (
(serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[10]s)
)
GROUP BY trace_id
HAVING t1_time > 0 AND t2_time > t1_time
) AS funnel
WHERE t2_time>t1_time
ORDER BY duration_ms DESC
LIMIT 5;`
LIMIT 5;
`
return fmt.Sprintf(queryTemplate,
containsErrorT1,
containsErrorT2,
@ -549,12 +548,13 @@ FROM (
(serviceName = step2.1 AND name = step2.2 AND (contains_error_t2 = 0 OR has_error = true) %[10]s)
)
GROUP BY trace_id
HAVING t1_time > 0 AND t2_time > t1_time
) AS funnel
WHERE
(t1_error = 1 OR t2_error = 1)
AND t2_time>t1_time
ORDER BY duration_ms DESC
LIMIT 5;`
LIMIT 5;
`
return fmt.Sprintf(queryTemplate,
containsErrorT1,
containsErrorT2,
@ -591,9 +591,34 @@ WITH
(%[6]d - %[5]d) / 1e9 AS time_window_sec,
('%[7]s', '%[8]s') AS step1,
('%[9]s','%[10]s') AS step2
('%[9]s', '%[10]s') AS step2,
, funnel AS (
%[1]d AS contains_error_t1,
%[2]d AS contains_error_t2
SELECT
round(total_s2_spans * 100.0 / total_s1_spans, 2) AS conversion_rate,
total_s2_spans / time_window_sec AS avg_rate,
greatest(sum_s1_error, sum_s2_error) AS errors,
avg_duration,
latency
FROM (
SELECT
countIf(t1_time > 0) AS total_s1_spans,
countIf(t1_time > 0 AND t2_time > t1_time) AS total_s2_spans,
sum(s1_error) AS sum_s1_error,
sum(s2_error) AS sum_s2_error,
avgIf(
(toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6,
t1_time > 0 AND t2_time > t1_time
) AS avg_duration,
quantileIf(%[13]s)(
(toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6,
t1_time > 0 AND t2_time > t1_time
) AS latency
FROM (
SELECT
trace_id,
minIf(timestamp, serviceName = step1.1 AND name = step1.2) AS t1_time,
@ -605,30 +630,12 @@ WITH
timestamp BETWEEN start_ts AND end_ts
AND serviceName IN (step1.1, step2.1)
AND name IN (step1.2, step2.2)
AND ((%[1]d = 0) OR (has_error AND serviceName = step1.1 AND name = step1.2))
AND ((%[2]d = 0) OR (has_error AND serviceName = step2.1 AND name = step2.2))
%[11]s
%[12]s
AND ((contains_error_t1 = 0) OR (has_error AND serviceName = step1.1 AND name = step1.2)) %[11]s
AND ((contains_error_t2 = 0) OR (has_error AND serviceName = step2.1 AND name = step2.2)) %[12]s
GROUP BY trace_id
)
, totals AS (
SELECT
countIf(t1_time > 0) AS total_s1_spans,
countIf(t1_time > 0 AND t2_time > t1_time) AS total_s2_spans,
greatest(sum(s1_error), sum(s2_error)) AS errors,
avg((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS avg_duration,
quantile(%[13]s)((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS latency
FROM funnel
)
SELECT
round(total_s2_spans * 100.0 / total_s1_spans, 2) AS conversion_rate,
total_s2_spans / time_window_sec AS avg_rate,
errors,
avg_duration,
latency
FROM totals;`
) AS funnel
) AS totals;
`
return fmt.Sprintf(tpl,
containsErrorT1,
@ -670,7 +677,7 @@ func BuildThreeStepFunnelStepOverviewQuery(
latencyTypeT2 string,
latencyTypeT3 string,
) string {
const baseTpl = `
const baseWithAndFunnel = `
WITH
toDateTime64(%[7]d/1e9, 9) AS start_ts,
toDateTime64(%[8]d/1e9, 9) AS end_ts,
@ -682,9 +689,9 @@ WITH
%[1]d AS contains_error_t1,
%[2]d AS contains_error_t2,
%[3]d AS contains_error_t3
%[3]d AS contains_error_t3,
, funnel AS (
funnel AS (
SELECT
trace_id,
minIf(timestamp, serviceName = step1.1 AND name = step1.2) AS t1_time,
@ -703,55 +710,62 @@ WITH
AND ((contains_error_t3 = 0) OR (has_error AND serviceName = step3.1 AND name = step3.2)) %[17]s
GROUP BY trace_id
)
, totals AS (
SELECT
countIf(t1_time > 0) AS total_s1_spans,
countIf(t1_time > 0 AND t2_time > t1_time) AS total_s2_spans,
countIf(t1_time > 0 AND t2_time > t1_time AND t3_time > t2_time) AS total_s3_spans,
sum(s1_error) AS sum_s1_error,
sum(s2_error) AS sum_s2_error,
sum(s3_error) AS sum_s3_error,
avg((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS avg_duration_12,
quantile(%[18]s)((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6) AS latency_12,
avg((toUnixTimestamp64Nano(t3_time) - toUnixTimestamp64Nano(t2_time)) / 1e6) AS avg_duration_23,
quantile(%[19]s)((toUnixTimestamp64Nano(t3_time) - toUnixTimestamp64Nano(t2_time)) / 1e6) AS latency_23
FROM funnel
)
`
const select12 = `
const totals12 = `
SELECT
round(total_s2_spans * 100.0 / total_s1_spans, 2) AS conversion_rate,
round(if(total_s1_spans > 0, total_s2_spans * 100.0 / total_s1_spans, 0), 2) AS conversion_rate,
total_s2_spans / time_window_sec AS avg_rate,
greatest(sum_s1_error, sum_s2_error) AS errors,
avg_duration_12 AS avg_duration,
latency_12 AS latency
FROM totals;
FROM (
SELECT
countIf(t1_time > 0 AND t2_time > t1_time) AS total_s2_spans,
countIf(t1_time > 0 AND t2_time > t1_time) AS total_s1_spans, -- eligible only
sum(s1_error) AS sum_s1_error,
sum(s2_error) AS sum_s2_error,
avgIf((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6, t1_time > 0 AND t2_time > t1_time) AS avg_duration_12,
quantileIf(%[18]s)((toUnixTimestamp64Nano(t2_time) - toUnixTimestamp64Nano(t1_time)) / 1e6, t1_time > 0 AND t2_time > t1_time) AS latency_12
FROM funnel
) AS totals;
`
const select23 = `
const totals23 = `
SELECT
round(total_s3_spans * 100.0 / total_s2_spans, 2) AS conversion_rate,
round(if(total_s2_spans > 0, total_s3_spans * 100.0 / total_s2_spans, 0), 2) AS conversion_rate,
total_s3_spans / time_window_sec AS avg_rate,
greatest(sum_s2_error, sum_s3_error) AS errors,
avg_duration_23 AS avg_duration,
latency_23 AS latency
FROM totals;
FROM (
SELECT
countIf(t2_time > 0 AND t3_time > t2_time) AS total_s3_spans,
countIf(t2_time > 0 AND t3_time > t2_time) AS total_s2_spans, -- eligible only
sum(s2_error) AS sum_s2_error,
sum(s3_error) AS sum_s3_error,
avgIf((toUnixTimestamp64Nano(t3_time) - toUnixTimestamp64Nano(t2_time)) / 1e6, t2_time > 0 AND t3_time > t2_time) AS avg_duration_23,
quantileIf(%[19]s)((toUnixTimestamp64Nano(t3_time) - toUnixTimestamp64Nano(t2_time)) / 1e6, t2_time > 0 AND t3_time > t2_time) AS latency_23
FROM funnel
) AS totals;
`
var selectTpl string
const fallback = `
SELECT 0 AS conversion_rate, 0 AS avg_rate, 0 AS errors, 0 AS avg_duration, 0 AS latency;
`
var totalsTpl string
switch {
case stepStart == 1 && stepEnd == 2:
selectTpl = select12
totalsTpl = totals12
case stepStart == 2 && stepEnd == 3:
selectTpl = select23
totalsTpl = totals23
default:
selectTpl = `SELECT 0 AS conversion_rate, 0 AS avg_rate, 0 AS errors, 0 AS avg_duration, 0 AS latency;`
totalsTpl = fallback
}
return fmt.Sprintf(
baseTpl+selectTpl,
baseWithAndFunnel+totalsTpl,
containsErrorT1,
containsErrorT2,
containsErrorT3,