> ## Documentation Index
> Fetch the complete documentation index at: https://docs.germainux.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Find Why Oracle E-Business Suite Concurrent Requests Pile Up or Fail

> Find out in minutes why EBS reports and jobs stay pending or end in error: a concurrent manager that stopped, a manager that blocks the others, or one program that fails on every run.

Components
[`Engine`](https://docs.germainux.com/deploy/germain-ux-engine)
[`RUM JS`](https://docs.germainux.com/deploy/germain-ux-rum-js)

> [card] [What GermainUX Watches](#what-germainux-watches)
>
> Managers, requests and changes

> [card] [Find the Cause](#find-the-cause)
>
> Three worked examples

> [card] [Alert and Act](#alert-and-act)
>
> Catch it next time

> [card] [Set It Up](#set-it-up)
>
> Queries, SLAs and alerts

## Overview

"My report has been pending for an hour." In Oracle E-Business Suite, invoices, interfaces, period-close jobs and reports all run as concurrent requests. When they stop running, finance and operations teams notice first, and the EBS team starts checking managers, queues and the database by hand.

A count of pending requests shows that something is wrong, not what. The cause can be a manager whose processes died, a manager that holds back every other one, or one program that fails each time it runs. Each one needs a different fix, and only one of them needs a restart.

GermainUX queries the concurrent processing tables on a schedule and keeps every manager, queue and request count as a KPI. When one moves, the analysis compares it with its baseline, shows the KPIs that moved with it, and lets you rule out the ones that stayed normal.

Alert
Aggregate Analysis
Related KPIs
Manager or Program
Fix and Confirm

This page uses the EBS concurrent processing tables as the example. The same approach works for any batch or job platform whose state GermainUX can query: schedulers, job tables and queues.

### Questions It Answers

* Why are users' requests stuck in Pending?
* Is a concurrent manager down, and which one?
* Are requests failing, or only waiting?
* Is the database available to the managers?
* Did last night's patch leave invalid objects behind?
* Is the backlog draining after the restart?

## What GermainUX Watches

GermainUX keeps each signal as a KPI, so the analysis can compare them with each other and with their own history.

| Signal | What it tells you | Where it comes from |
| - | - | - |
| **Manager processes** | How many processes each manager runs: Standard, Internal and Conflict Resolution | [EBS KPIs](https://docs.germainux.com/integrations/oracle-e-business-suite-monitoring.md#kpis) |
| **Pending and running requests** | Whether requests wait, and whether the managers pick them up | [EBS KPIs](https://docs.germainux.com/integrations/oracle-e-business-suite-monitoring.md#kpis) |
| **Requests with errors and warnings** | How many requests end in error or warning, and when the count started to climb | [EBS KPIs](https://docs.germainux.com/integrations/oracle-e-business-suite-monitoring.md#kpis) |
| **Changes** | Invalid database objects and newly registered concurrent programs after a patch | [EBS KPIs](https://docs.germainux.com/integrations/oracle-e-business-suite-monitoring.md#kpis) |
| **Database availability** | Whether the EBS database answers the monitoring queries at all | [Oracle DB Observability](https://docs.germainux.com/integrations/oracle-db-observability.md) |
| **The rest of the landscape** | The browser UIs users work in, the application servers, logs and hosts | [Browser monitoring](https://docs.germainux.com/integrations/browser-monitoring.md), [Oracle WebLogic](https://docs.germainux.com/integrations/oracle-weblogic-monitoring.md), [logs](https://docs.germainux.com/integrations/file-log-monitoring.md) |

## Find the Cause

The investigation follows the same path every time:

1. Open [Aggregate Analysis](https://docs.germainux.com/dashboards/analytics-dashboard/4-analysis-dashboard.md#aggregate-view-vs-instance-view) on the KPI that raised the alert, for the period of the problem. GermainUX compares it with the period before.
2. Read the [Relationships](https://docs.germainux.com/dashboards/analytics-dashboard/4-analysis-dashboard.md#relationships) chart and the [Related KPIs](https://docs.germainux.com/dashboards/analytics-dashboard/4-analysis-dashboard.md#related-kpis). Under **Related Metrics**, GermainUX lists the manager, request and availability KPIs that moved at the same time. Click one to add it to the chart.
3. Rule out what stayed normal: a related KPI that didn't move isn't listed. It tells the EBS team where not to look.
4. Fix the cause, then run the same analysis on the period after the fix.

The three examples below come from simulated EBS data, and their numbers belong to the example. They start from similar symptoms and end at different answers.

### Example: The Standard Manager Goes Down

Mid-morning, the Standard Manager's processes die. It runs most programs, so new requests start to wait. In this example, pending requests climb from their usual 140 to about 450 in 90 minutes.

**1. Spot the climb**

Open the analysis of **EBS Query Pending Concurrent Requests** for the morning. Most of the usual 140 pending requests are scheduled ones waiting for their start time, so a steady level is normal. A climb is not.

In this example, the average for the morning is 177.738, up 26.1% on 140.973 for the same hours on the previous days. The Relationships chart shows hourly averages, so the peak reads about 300 rather than the highest poll. The second peak, on the afternoon of 12 November, is the next example.

![Analysis of EBS Query Pending Concurrent Requests for the morning of 11 November: an average of 177.738 against 140.973 on the previous days, up 26.1%. The Relationships chart shows hourly pending requests near 140 that peak at about 300 on the morning of 11 November and again on the afternoon of 12 November. Under Related Metrics, a card for the active processes of the Standard Manager queue shows 7.857 against a 10.000 baseline, down 21%](https://mintcdn.com/germain-ux/0hQY3qzJj2OStUPN/images/main/ebs-concurrent-request-analysis/smq-down-pending-requests-climb.png?fit=max\&auto=format\&n=0hQY3qzJj2OStUPN\&q=85\&s=7789c3b17ec69a10ee6e1a9503543351)

Pending Concurrent Requests Climb on the Morning of 11 November - GermainUX

**2. Check the managers**

Under **Related Metrics**, GermainUX lists the related KPIs that moved in the same period. Click the card of the Standard Manager queue's active processes to add it to the Relationships chart, and check whether they fall to 0 when the climb starts. In this example, the Standard Manager runs 10 processes, and its average for the morning is 7.857, down 21% on 10.000.

![Analysis of EBS Query Pending Concurrent Requests for the morning of 11 November: an average of 177.738 against 140.973 on the previous days, up 26.1%. The Relationships chart shows hourly pending requests near 140 that peak at about 300 on the morning of 11 November, while the active processes of the Standard Manager queue fall from 10 to 0 at the same time. Under Related Metrics, the Standard Manager queue card shows 7.857 against a 10.000 baseline, down 21%](https://mintcdn.com/germain-ux/0hQY3qzJj2OStUPN/images/main/oracle-e-business-suite-monitoring/ebs-pending-requests-analysis.png?fit=max\&auto=format\&n=0hQY3qzJj2OStUPN\&q=85\&s=4dddda4b445143cd8710e3f2dc62fbb1)

Pending Requests Climb While the Standard Manager Is Down - GermainUX

**3. Rule out the rest**

The Internal and Conflict Resolution Managers keep their processes, the database stays available, and the error and warning counts don't move, so none of them shows under **Related Metrics**. Requests aren't failing; nothing runs them.

Restart the Standard Manager. Its processes come back to 10, running requests jump while the backlog clears, and pending requests are back at 140 within two polls.

### Example: The Conflict Resolution Manager Blocks Every Request

The next afternoon, pending requests climb again, faster this time: from 140 to about 580 in 80 minutes. The Standard Manager looks healthy.

**1. Spot the climb**

Open the analysis of **EBS Query Pending Concurrent Requests** for the afternoon. Every new request stays pending.

In this example, the average for the afternoon is 199.736, up 41.5% on 141.153 for the same hours on the previous days. The hourly chart peaks at about 330, not 580, because it shows hourly averages.

![Analysis of EBS Query Pending Concurrent Requests for the afternoon of 12 November: an average of 199.736 against 141.153 on the previous days, up 41.5%. The Relationships chart shows hourly pending requests near 140 that peak at about 330 on the afternoon of 12 November. Under Related Metrics, a card for the active processes of the Conflict Resolution Manager queue shows 0.778 against a 1.000 baseline, down 22%](https://mintcdn.com/germain-ux/0hQY3qzJj2OStUPN/images/main/ebs-concurrent-request-analysis/crm-down-pending-requests-climb.png?fit=max\&auto=format\&n=0hQY3qzJj2OStUPN\&q=85\&s=b30b1835dfc3e09d99e635d6d059596f)

Pending Concurrent Requests Climb on the Afternoon of 12 November - GermainUX

**2. Compare running and pending**

Open the analysis of **EBS Query Running Concurrent Requests** for the same afternoon. Running requests normally peak late in the morning. Look for running requests that drop to almost 0 while pending requests climb: under **Related Metrics**, the pending requests card shows the climb. The Standard Manager queue isn't listed, because it keeps all its processes. The managers are up, but idle.

![Analysis of EBS Query Running Concurrent Requests for the afternoon of 12 November: an average of 4.111 against 4.638 on the previous days, down 11.4%. The Relationships chart shows hourly running requests that peak at about 7 late each morning and drop to almost 0 at 2 pm on 12 November, with the active processes of the Conflict Resolution Manager queue dipping from 1 to 0 at the same time. Under Related Metrics, pending requests show 199.736 against a 141.153 baseline, up 42%, and the Conflict Resolution Manager queue shows 0.778 against 1.000, down 22%](https://mintcdn.com/germain-ux/0hQY3qzJj2OStUPN/images/main/ebs-concurrent-request-analysis/crm-down-running-requests-analysis.png?fit=max\&auto=format\&n=0hQY3qzJj2OStUPN\&q=85\&s=d11a6b4080ac504d68329cb3eb1fd2fb)

Running Requests Drain While Pending Requests Climb - GermainUX

**3. Find the manager that holds them back**

Go back to the analysis of pending requests. Under **Related Metrics**, click the card of the Conflict Resolution Manager queue's active processes to add it to the Relationships chart, and check whether they fall to 0 when the climb starts. Without it, no request is released to the other managers. In this example, the Conflict Resolution Manager runs 1 process.

![Analysis of EBS Query Pending Concurrent Requests for the afternoon of 12 November: an average of 199.736 against 141.153 on the previous days, up 41.5%. The Relationships chart shows hourly pending requests near 140 that peak at about 330 on the afternoon of 12 November, while the active processes of the Conflict Resolution Manager queue fall from 1 to 0 at the same time. Under Related Metrics, the Conflict Resolution Manager queue card shows 0.778 against a 1.000 baseline, down 22%](https://mintcdn.com/germain-ux/0hQY3qzJj2OStUPN/images/main/ebs-concurrent-request-analysis/crm-down-pending-requests-analysis.png?fit=max\&auto=format\&n=0hQY3qzJj2OStUPN\&q=85\&s=bafd51fda6694e7df75607a0173134d2)

Pending Requests Climb While the Conflict Resolution Manager Is Down - GermainUX

**4. Rule out the rest**

The Standard and Internal Managers keep their processes, the database stays available, and no request ends in error, so none of them shows under **Related Metrics**.

Restarting the Standard Manager wouldn't help here. Restart the Conflict Resolution Manager, then check that running requests return and the backlog drains.

### Example: One Interface Import Fails on Every Run

From 10:00, the number of requests with errors starts to grow. Nobody complains about pending requests, because nothing waits.

**1. Spot the growth**

Open the analysis of **EBS Query Concurrent Requests With Error** for the morning. In this example, the count grows by 60 in two hours, from about 300 to about 360, against about 6 on a normal morning. The average for the morning is 324.069, up 15.0% on 281.858 for the same hours on the previous days. Requests with warnings grow by the same 60.

![Analysis of EBS Query Concurrent Requests With Error for the morning of 15 November: an average of 324.069 against 281.858 on the previous days, up 15.0%, and a Relationships chart where the count steps up from about 300 to about 360 between 10:00 and noon, then stays at the new level](https://mintcdn.com/germain-ux/0hQY3qzJj2OStUPN/images/main/ebs-concurrent-request-analysis/import-fails-error-requests-analysis.png?fit=max\&auto=format\&n=0hQY3qzJj2OStUPN\&q=85\&s=c253738665850315e64803c37c68a8c5)

Requests With Errors Grow While One Interface Import Fails - GermainUX

**2. Read the pattern**

The count rises in a straight line, not a jump, and stays at the new level. In this example, both counts grow by one every two minutes. A steady, regular step points to one program that is resubmitted on a schedule and fails each time, not to the platform.

**3. Rule out the rest**

Pending and running requests, the manager processes and database availability all stay at their baseline. The managers run requests normally; this one fails.

This answer goes to the owner of the interface, not the EBS administrators. Open the requests that ended in error since 10:00 in EBS to find the program, fix its data or setup, and check that the error count stops growing. The count stays high until the requests are purged, so watch its growth, not its level.

## Alert and Act

Each example points to a signal you can watch from now on:

* **[SLAs](https://docs.germainux.com/analytics/slas.md)** on the active processes of each manager queue, so a manager that stops raises an alert before the backlog grows
* **[SLAs](https://docs.germainux.com/analytics/slas.md)** on pending requests and on the growth of requests with errors, with [outlier detection](https://docs.germainux.com/analytics/outlier-detection.md) for the times of day when the normal level changes
* **[Alerts](https://docs.germainux.com/automation/notifications/create-an-alert.md)** to the EBS team by email, SMS, [Microsoft Teams](https://docs.germainux.com/automation/notifications/microsoft-teams-notification.md) or [Slack](https://docs.germainux.com/automation/notifications/slack.md), and incidents in [ServiceNow](https://docs.germainux.com/automation/servicenow.md)
* **[Automated actions](https://docs.germainux.com/automation/automated-actions.md)** that run an approved script, for example to restart a manager that stopped
* **An alert on invalid objects**, so a patch that leaves objects invalid is recompiled before the morning's requests run

## Set It Up

**1. Monitor the concurrent processing tables**

[Deploy EBS monitoring](https://docs.germainux.com/integrations/oracle-e-business-suite-monitoring.md#deploy) on a GermainUX Engine that can reach the EBS database. It queries the managers, requests, programs and invalid objects on a schedule.

**2. Monitor the database**

Add [Oracle Database monitoring](https://docs.germainux.com/integrations/oracle-db-observability.md) for the EBS database, so the analysis can rule database problems in or out.

**3. Add your own queries**

When you need more detail, such as errors per program, add a [custom query](https://docs.germainux.com/integrations/oracle-db-observability/configure-monitoring-for-oracle-database.md) with the program name as a dimension.

**4. Set SLAs and alerts**

Set [SLAs](https://docs.germainux.com/analytics/slas/configure-sla.md) on the manager, pending-request and error KPIs, and [alerts](https://docs.germainux.com/automation/notifications/create-an-alert.md) to the team that owns each one.

## Additional Resources

> [card] [Oracle E-Business Suite Monitoring](https://docs.germainux.com/integrations/oracle-e-business-suite-monitoring.md)
>
> Concurrent processing KPIs

> [card] [Analysis Dashboard](https://docs.germainux.com/dashboards/analytics-dashboard/4-analysis-dashboard.md)
>
> Aggregate and instance analysis

> [card] [Use Cases](https://docs.germainux.com/use-cases/overview.md)
>
> More Use Cases

## Get Help

Every EBS instance has its own managers, schedules and critical programs. The GermainUX team can help you choose the queues and programs to watch, set their SLAs and route the alerts to the right team.

> [card] [Contact GermainUX Support](https://docs.germainux.com/getting-started/support.md)
>
> The GermainUX team helps you get up and running faster, empower your team, turn data into actionable insights, and optimize performance to deliver measurable business outcomes.

> [card] Can't Find What You Need?
>
> Search all GermainUX docs with **Ctrl + K**, or use the feedback buttons at the bottom of this page to tell us what is missing.

The content of this document is proprietary to GERMAIN SOFTWARE.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.